I just want to make sure of something: are the EOMA68-A20 computer cards going to be shipped with logging and journaling disabled (so that the storage isn't constantly being written to)? I'm asking because this is pretty much a necessity for an OS running on NAND or an SD card if you don't want to have to constantly buy new SD cards because the previous one reached its write limit.
Ditto for swap, but given the small amount of NAND on the A20 card, I'm sure not having swap is a given anyway. ;)
2017-02-08 15:33 GMT+01:00 Julie Marchant onpon4@riseup.net:
I just want to make sure of something: are the EOMA68-A20 computer cards going to be shipped with logging and journaling disabled (so that the storage isn't constantly being written to)? I'm asking because this is pretty much a necessity for an OS running on NAND or an SD card if you don't want to have to constantly buy new SD cards because the previous one reached its write limit.
NAND wear is probably more linked to age than number of writes. NAND has more faults than harddisk. The trick is ECC and headroom and mapping for faulted sectors.
The biggest issue is boot fixed read addresses.
Ditto for swap, but given the small amount of NAND on the A20 card, I'm sure not having swap is a given anyway. ;)
-- Julie Marchant https://onpon4.github.io
Protect your emails with GnuPG: https://emailselfdefense.fsf.org
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
On 02/08/2017 09:46 AM, mike.valk@gmail.com wrote:
NAND wear is probably more linked to age than number of writes.
No, NAND is flash memory, it has a definite limit on the number of times it can be written to (in each sector, that is). It doesn't matter how old flash memory is, if you reach its write limit, it's useless.
All flash memory is like this, including most SSDs. The only variation is what the limit is and what the firmware does to compensate. So one might take more writes than another, but regardless, every write *necessarily* brings it closer to the end of its life.
Op 8 feb. 2017 16:11 schreef "Julie Marchant" onpon4@riseup.net:
On 02/08/2017 09:46 AM, mike.valk@gmail.com wrote:
NAND wear is probably more linked to age than number of writes.
No, NAND is flash memory, it has a definite limit on the number of times it can be written to (in each sector, that is). It doesn't matter how old flash memory is, if you reach its write limit, it's useless.
All flash memory is like this, including most SSDs. The only variation is what the limit is and what the firmware does to compensate. So one might take more writes than another, but regardless, every write *necessarily* brings it closer to the end of its life.
.
That's debatable: https://www.google.nl/amp/embedded#amp=http%253A%252F%252Fwww.zdnet.com%252F...
-- Julie Marchant https://onpon4.github.io
Protect your emails with GnuPG: https://emailselfdefense.fsf.org
_______________________________________________ arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
On Wed, Feb 8, 2017 at 2:54 PM, mike.valk@gmail.com mike.valk@gmail.com wrote:
Op 8 feb. 2017 16:11 schreef "Julie Marchant" onpon4@riseup.net:
On 02/08/2017 09:46 AM, mike.valk@gmail.com wrote:
NAND wear is probably more linked to age than number of writes.
No, NAND is flash memory, it has a definite limit on the number of times it can be written to (in each sector, that is). It doesn't matter how old flash memory is, if you reach its write limit, it's useless.
All flash memory is like this, including most SSDs. The only variation is what the limit is and what the firmware does to compensate. So one might take more writes than another, but regardless, every write *necessarily* brings it closer to the end of its life.
.
That's debatable: https://www.google.nl/amp/embedded#amp=http%253A%252F% 252Fwww.zdnet.com%252Fgoogle-amp%252Farticle%252Fssd- reliability-in-the-real-world-googles-experience%252F&idx=0
Link doesn't work for me (Thanks Google!) but this one does.
http://www.zdnet.com/article/ssd-reliability-in-the-real-world-googles-exper...
Assuming this is what you meant to link to.
-- Julie Marchant https://onpon4.github.io
Protect your emails with GnuPG: https://emailselfdefense.fsf.org
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
On 02/08/17 15:02, Adam Van Ymeren wrote:
On Wed, Feb 8, 2017 at 2:54 PM, mike.valk@gmail.com That's debatable: https://www.google.nl/amp/embedded#amp=http%253A%252F%252Fwww.zdnet.com%252F... https://www.google.nl/amp/embedded#amp=http%253A%252F%252Fwww.zdnet.com%252Fgoogle-amp%252Farticle%252Fssd-reliability-in-the-real-world-googles-experience%252F&idx=0
Link doesn't work for me (Thanks Google!) but this one does.
http://www.zdnet.com/article/ssd-reliability-in-the-real-world-googles-exper...
Assuming this is what you meant to link to.
Thank you.
Mike, if that article is accurate, the study doesn't contradict what I said:
None of the drives in the study came anywhere near their write limits, even the 3,000 writes specified for the MLC drives
On 02/08/2017 07:15 PM, Julie Marchant wrote:
Mike, if that article is accurate, the study doesn't contradict what I said:
None of the drives in the study came anywhere near their write limits, even the 3,000 writes specified for the MLC drives
I wrote this while I was at work on my break, so now that I have more time, let me elaborate.
Supposedly, in all cases tested, none of the SSDs were written to anywhere near enough times to cause substantial wear. It is mentioned in this article summarizing it that even 3,000 writes were not reached. Most flash media can sustain at least tens of thousands of writes, so if not even 3,000 writes were reached, it makes sense that none of the SSDs tested failed.
However, it doesn't follow that this kind of wear is insignificant for all flash memory. In particular:
* SSDs tend to be larger than SD cards, so they're not going to wear out as fast unless the entirety of the space is being used and constantly changed. * The firmware of SSDs may be different from the firmware of SD cards. For that matter, the firmware of SSDs may be different from the firmware of other SSDs. Don't underestimate the power of wear leveling.
It still holds true that every change to any flash memory brings it closer to the end of its life. It's just that other factors can do this, too, and in some cases (e.g. SSDs) it may mean that the write cycle limit is less important in practice.
On Wed, Feb 08, 2017 at 11:20:32PM -0500, Julie Marchant wrote:
On 02/08/2017 07:15 PM, Julie Marchant wrote:
Mike, if that article is accurate, the study doesn't contradict what I said:
None of the drives in the study came anywhere near their write limits, even the 3,000 writes specified for the MLC drives
I wrote this while I was at work on my break, so now that I have more time, let me elaborate.
Supposedly, in all cases tested, none of the SSDs were written to anywhere near enough times to cause substantial wear. It is mentioned in this article summarizing it that even 3,000 writes were not reached. Most flash media can sustain at least tens of thousands of writes, so if not even 3,000 writes were reached, it makes sense that none of the SSDs tested failed.
However, it doesn't follow that this kind of wear is insignificant for all flash memory. In particular:
- SSDs tend to be larger than SD cards, so they're not going to wear out
as fast unless the entirety of the space is being used and constantly changed.
- The firmware of SSDs may be different from the firmware of SD cards.
For that matter, the firmware of SSDs may be different from the firmware of other SSDs. Don't underestimate the power of wear leveling.
It still holds true that every change to any flash memory brings it closer to the end of its life. It's just that other factors can do this, too, and in some cases (e.g. SSDs) it may mean that the write cycle limit is less important in practice.
A test we did for our systems is to constantly (re)write and sync a large random file for many cycles (enough to simulate several years of normal operation), and generally encountered hardly any faults.
Feel free to test this yourself.
On 02/09/2017 05:22 AM, Tzafrir Cohen wrote:
A test we did for our systems is to constantly (re)write and sync a large random file for many cycles (enough to simulate several years of normal operation), and generally encountered hardly any faults.
Feel free to test this yourself.
Do your systems run on microSD cards? Because that is after all what matters here. Like I said, while it's *fundamentally* the same for both, SSDs are more likely to have more storage space and more sophisticated firmware compared to microSD cards because they're typically used for completely different purposes.
On Thu, Feb 09, 2017 at 06:57:48AM -0500, Julie Marchant wrote:
On 02/09/2017 05:22 AM, Tzafrir Cohen wrote:
A test we did for our systems is to constantly (re)write and sync a large random file for many cycles (enough to simulate several years of normal operation), and generally encountered hardly any faults.
Feel free to test this yourself.
Do your systems run on microSD cards?
Yes (8GB or 32GB).
On 02/09/2017 07:03 AM, Tzafrir Cohen wrote:
Yes (8GB or 32GB).
That's good to hear.
I've had an SD card that became unusable (as in it was literally impossible to write to successfully, even impossible just to format) after Pandian (a version of Debian for the OpenPandora) was on it for some time. You seem to be knowledgeable about this subject; do you have any idea what could cause all attempts at writing to an SD card to be unsuccessful, as if it was a ROM? I've always assumed that it had been written to too much.
2017-02-08 21:02 GMT+01:00 Adam Van Ymeren adam.vany@gmail.com:
On Wed, Feb 8, 2017 at 2:54 PM, mike.valk@gmail.com mike.valk@gmail.com wrote:
Op 8 feb. 2017 16:11 schreef "Julie Marchant" onpon4@riseup.net:
On 02/08/2017 09:46 AM, mike.valk@gmail.com wrote:
NAND wear is probably more linked to age than number of writes.
No, NAND is flash memory, it has a definite limit on the number of times it can be written to (in each sector, that is). It doesn't matter how old flash memory is, if you reach its write limit, it's useless.
All flash memory is like this, including most SSDs. The only variation is what the limit is and what the firmware does to compensate. So one might take more writes than another, but regardless, every write *necessarily* brings it closer to the end of its life.
.
That's debatable: https://www.google.nl/amp/embedded#amp=http%253A%252F%252Fww w.zdnet.com%252Fgoogle-amp%252Farticle%252Fssd-reliabilit y-in-the-real-world-googles-experience%252F&idx=0
Link doesn't work for me (Thanks Google!) but this one does.
http://www.zdnet.com/article/ssd-reliability-in-the-real- world-googles-experience/
Assuming this is what you meant to link to.
Yes thank you
-- Julie Marchant https://onpon4.github.io
Protect your emails with GnuPG: https://emailselfdefense.fsf.org
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
On Wed, Feb 08, 2017 at 09:33:04AM -0500, Julie Marchant wrote:
I just want to make sure of something: are the EOMA68-A20 computer cards going to be shipped with logging and journaling disabled (so that the storage isn't constantly being written to)? I'm asking because this is pretty much a necessity for an OS running on NAND or an SD card if you don't want to have to constantly buy new SD cards because the previous one reached its write limit.
Ditto for swap, but given the small amount of NAND on the A20 card, I'm sure not having swap is a given anyway. ;)
Just one data point from systems we have, based on Raspberry Pi2 (SD card):
* By default we disable all standard logging to disk. - It is possible to reenable logging to (r)syslog - and/or create /var/log/journal (We keep Debian's default of Storage=auto). * We tried working without a swap. After some crashes we realised it was needed.
On 02/08/2017 09:48 AM, Tzafrir Cohen wrote:
- We tried working without a swap. After some crashes we realised it was needed.
With 1 GB of RAM? You must be running the wrong programs, then. The original OpenPandora, with only 256 MB of RAM, works perfectly fine without swap in my experience. You just need to be mindful of how many programs you're running at once. I seriously doubt people are commonly going to run into problems with 2GB of RAM.
Besides, swap isn't a solution to low RAM, it's a failsafe. Some people might find the cost of degrading a microSD card to be worth that failsafe, as opposed to just letting some programs crash, but I think most people would prefer the latter option for the most part.
On Feb 8, 2017 10:19 AM, "Julie Marchant" onpon4@riseup.net wrote:
Besides, swap isn't a solution to low RAM, it's a failsafe. Some people might find the cost of degrading a microSD card to be worth that failsafe, as opposed to just letting some programs crash, but I think most people would prefer the latter option for the most part.
But if you really use it as a failsafe (that is, very rarely, and only long enough to notice the slowdown and close some programs), it doesn't degrade the flash much at all. It's only, IMO, a big deal when you're trying to pretend your system has more RAM than it does, and thus using swap regularly.
As you say, different people will make different choices, but for me swap-on-SD is clearly worth it, so if/when an out-of-memory condition _does_ occur, I get a chance to choose which processes live and die, and to make sure any important work is saved properly. I don't wholly distrust the kernel OOM killer's heuristics, but heuristics are never perfect, and the cost in flash usage seems very reasonable to me.
However, since the default needs to be one or the other, and we can't even know that users will _have_ a microSD card in it, I'd agree that the factory image should have swap disabled. But I'd say it should have documentation or scripts to make it easy for end users to enable swap on removable (thus replaceable) storage if they want to. And it _definitely_ shouldn't ship with a kernel or userland incapable of swapping, like some Android devices I've used; but I expect we all agree on that.
Benson Mitchell
On Wed, 08 Feb 2017 11:27:55 -0500, Benson Mitchell wrote:
As you say, different people will make different choices, but for me swap-on-SD is clearly worth it, so if/when an out-of-memory condition _does_ occur, I get a chance to choose which processes live and die, and to make sure any important work is saved properly. I don't wholly distrust the kernel OOM killer's heuristics, but heuristics are never perfect, and the cost in flash usage seems very reasonable to me.
My experience when I end up swapping to a hard drive is that the whole system becomes so slow and unresponsive that it mosstsly ignores console input, and I have no ability to kill things before the OOM does. Often the best way to get things moving again is a hard reset. And then watching it repair the filesystems.
-- hendrik
--- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Wed, Feb 8, 2017 at 2:33 PM, Julie Marchant onpon4@riseup.net wrote:
I just want to make sure of something: are the EOMA68-A20 computer cards going to be shipped with logging and journaling disabled (so that the storage isn't constantly being written to)? I'm asking because this is pretty much a necessity for an OS running on NAND or an SD card if you don't want to have to constantly buy new SD cards because the previous one reached its write limit.
64bit somethingorother is off. i'm giving serious consideration to just putting the OS on a small microsd card and shipping that out.
i really wanted to put ubifs on but it's not working well with 3.4
Ditto for swap, but given the small amount of NAND on the A20 card, I'm sure not having swap is a given anyway. ;)
yeah no on the swap
On 02/08/2017 11:37 AM, Luke Kenneth Casson Leighton wrote:
64bit somethingorother is off. i'm giving serious consideration to just putting the OS on a small microsd card and shipping that out.
i really wanted to put ubifs on but it's not working well with 3.4
That's not really what I was asking. Regardless of what storage device it's on, I just want to make sure the filesystem doesn't use journaling (e.g. ext2 rather than ext4) and that the system is set up to not record logs of things like warnings and errors. These things will do a number on an SD card. (I can't say I know what you're referring to by "64bit somethingorother".)
It won't hurt me personally either way because I'm capable of reformatting to a non-journaling FS and should be capable of disabling logging myself, but it would suck if your computer cards got a reputation for chewing through SD cards because of the default settings.
--- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Wed, Feb 8, 2017 at 6:35 PM, Julie Marchant onpon4@riseup.net wrote:
On 02/08/2017 11:37 AM, Luke Kenneth Casson Leighton wrote:
64bit somethingorother is off. i'm giving serious consideration to just putting the OS on a small microsd card and shipping that out.
i really wanted to put ubifs on but it's not working well with 3.4
That's not really what I was asking. Regardless of what storage device it's on, I just want to make sure the filesystem doesn't use journaling (e.g. ext2 rather than ext4) and that the system is set up to not record logs of things like warnings and errors. These things will do a number on an SD card. (I can't say I know what you're referring to by "64bit somethingorother".)
i'm saying i have to switch off a couple of flags anyway so that should remind me to remove journalling as well.
l.
It won't hurt me personally either way because I'm capable of reformatting to a non-journaling FS and should be capable of disabling logging myself, but it would suck if your computer cards got a reputation for chewing through SD cards because of the default settings.
appreciated. thx julie.
On 02/08/2017 02:00 PM, Luke Kenneth Casson Leighton wrote:
i'm saying i have to switch off a couple of flags anyway so that should remind me to remove journalling as well.
Alright, that's good to hear. :) Just don't forget the logging bit, as well; that's a separate thing from journaling and equally important. (Journaling is something the filesystem does, logging is something either syslogd or rsyslog does.)
--- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Wed, Feb 8, 2017 at 7:34 PM, Julie Marchant onpon4@riseup.net wrote:
On 02/08/2017 02:00 PM, Luke Kenneth Casson Leighton wrote:
i'm saying i have to switch off a couple of flags anyway so that should remind me to remove journalling as well.
Alright, that's good to hear. :) Just don't forget the logging bit, as well; that's a separate thing from journaling and equally important. (Journaling is something the filesystem does, logging is something either syslogd or rsyslog does.)
do you know the mkfs.ext4 commands needed or the ext2flags command?
On Thu, Feb 09, 2017 at 06:03:06AM -0500, Julie Marchant wrote:
On 02/09/17 01:50, Luke Kenneth Casson Leighton wrote:
do you know the mkfs.ext4 commands needed or the ext2flags command?
No, but just using an ext2 filesystem instead should suffice (since ext2 doesn't support journaling). That's the standard advice for the OpenPandora.
So what happens on a violent poweroff? Just lose data?
There's also f2fs. But you do need a recent enough kernel to use it. The versions from some older kernels may eat some files occasionally.
2017-02-09 12:03 GMT+01:00 Julie Marchant onpon4@riseup.net:
On 02/09/17 01:50, Luke Kenneth Casson Leighton wrote:
do you know the mkfs.ext4 commands needed or the ext2flags command?
No, but just using an ext2 filesystem instead should suffice (since ext2 doesn't support journaling). That's the standard advice for the OpenPandora.
As we all have established. We have flash memory. Which comes in a few flavors. NAND (Not AND) and NOR (Not Or). Which is the type of logic gate used.
Then we have SLC (Single level), MLC (Multi level) (And TLC, QLC)
With MLC, TLC, QLC you read/write multiple bits at once. These come with an issue: https://en.wikipedia.org/wiki/Solid-state_drive#Battery_or_supercapacitor
Then we have wear leveling. With bare Flash you have none. With SD/MCC/eMMC and SSD you have some. This is done via hardware/firmware. This is useally done in a way that is support regular rotating disk filesystem.
The mechanics are though obscure and those controller usually are programmed by closed firmware.
So for bare NAND you should not use a regular filesystem like EXT, XFS, FAT, NTFS, etc. Also F2FS is not build for bare nand. It is made for controlled flash.
For bare NAND you should use special drivers and filesystems like MTD and UBIFS or YaFFS. http://linux-sunxi.org/NAND
The Allwinner 3.4 kernel contains a NAND driver. But I don't know what the quality of it is. I though that I read it was not that great and that the new mainline driver had a lot of hoops to go trough to support flash formatted/controlled by the 3.4 driver.
So MLC has problems with power loss. So a journalling filesystem is a must. A journalling FS has noting to do with logging. But with write consistency.
With (e)MMC/SD adn SSD you lose a bit of freedom and gain extra security issues. But makes programming easier.
So weather NAND suffers from wear or age: You need a good controller (hardware or software) identifying broken cells and marking them unfit and not bother progams or users with unwritable sectors.
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
2017-02-10 8:53 GMT+01:00 mike.valk@gmail.com mike.valk@gmail.com:
2017-02-09 12:03 GMT+01:00 Julie Marchant onpon4@riseup.net:
On 02/09/17 01:50, Luke Kenneth Casson Leighton wrote:
do you know the mkfs.ext4 commands needed or the ext2flags command?
No, but just using an ext2 filesystem instead should suffice (since ext2 doesn't support journaling). That's the standard advice for the OpenPandora.
As we all have established. We have flash memory. Which comes in a few flavors. NAND (Not AND) and NOR (Not Or). Which is the type of logic gate used.
Then we have SLC (Single level), MLC (Multi level) (And TLC, QLC)
With MLC, TLC, QLC you read/write multiple bits at once. These come with an issue: https://en.wikipedia.org/wiki/Solid-state_drive#Battery_or_supercapacitor
Then we have wear leveling. With bare Flash you have none. With SD/MCC/eMMC and SSD you have some. This is done via hardware/firmware. This is useally done in a way that is support regular rotating disk filesystem.
The mechanics are though obscure and those controller usually are programmed by closed firmware.
So for bare NAND you should not use a regular filesystem like EXT, XFS, FAT, NTFS, etc. Also F2FS is not build for bare nand. It is made for controlled flash.
For bare NAND you should use special drivers and filesystems like MTD and UBIFS or YaFFS. http://linux-sunxi.org/NAND
The Allwinner 3.4 kernel contains a NAND driver. But I don't know what the quality of it is. I though that I read it was not that great and that the new mainline driver had a lot of hoops to go trough to support flash formatted/controlled by the 3.4 driver.
So MLC has problems with power loss. So a journalling filesystem is a must. A journalling FS has noting to do with logging. But with write consistency.
The biggest writer in a desktopish environment are the system and application logs.
So move var log to ram. This is usually enough for debugging purposes. But it should remain easily possible to write to flash/disk if debugging hanging systems issues.
Further the files system "acces time" stamps is a big writer. With BTRFS you have the "relatime" option (relax acces time updates).
This does break some programs which rely on accurate "atime". But for most end users this is not a problem.
With (e)MMC/SD adn SSD you lose a bit of freedom and gain extra security issues. But makes programming easier.
So weather NAND suffers from wear or age: You need a good controller (hardware or software) identifying broken cells and marking them unfit and not bother progams or users with unwritable sectors.
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
On Friday 10 Feb 2017 15:29:53 mike.valk@gmail.com wrote:
o move var log to ram. This is usually enough for debugging purposes. But it should remain easily possible to write to flash/disk if debugging hanging systems issues.
Further the files system "acces time" stamps is a big writer. With BTRFS you have the "relatime" option (relax acces time updates).
noatime with other fs.
This does break some programs which rely on accurate "atime". But for most end users this is not a problem.
works flawlessly for my use case with debian.
I just want to make sure of something: are the EOMA68-A20 computer cards going to be shipped with logging and journaling disabled (so that the storage isn't constantly being written to)? I'm asking because this is
FWIW, I think that journaling is a non-issue. The underlying assumption is that journaling causes more writes and hence ages the NAND faster. While this is true in some cases, it's not true in all cases (the reverse can also be true in some cases), and whether it's true in the common case is still up for debate AFAIK. Furthermore, even if when it's true, the factor by which it is increased is small (IIRC the worst case is a factor 2).
Now it's true that all those extra writes go to the same place on the filesystem, so if your NAND's wear leveling sucks rocks, you'll kill your NAND very quickly. But if your NAND's wear-leveling is anywhere close to honest, then journalling shouldn't make any noticeable difference to the expected lifetime of your NAND.
There's been a few reports of people seeing that SD die after a fairly short use of a journaling filesystem on it. I don't think I've ever seen conclusive evidence that the two were related (e.g. I've also seen SD cards die even though they've never touched a journalling filesystem), and I've seen lots of SD cards live long happy lifes with a journalling filesystem on them.
So, IMO, go ahead and take advantage of the latest journalling filesystem. If you SD card dies, it's just that you got a lemon.
W.r.t to logging, I've agree that you're probably better off logging to RAM (or to a remote host) than to a local "disk", and AFAIK that's the default behavior of systemd anyway.
Stefan
--- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Thu, Feb 9, 2017 at 1:50 PM, Stefan Monnier monnier@iro.umontreal.ca wrote:
W.r.t to logging, I've agree that you're probably better off logging to RAM (or to a remote host) than to a local "disk", and AFAIK that's the default behavior of systemd anyway.
with the exception of fedora which has only a few backers i will NOT be distributing a filesystem which contains the completely unethically-developed and very dangerous systemd application. having evaluated its development, watched the predicted security vulnerabilities unfold and cause massive disruption, and witnessed its "ram it down people's throats" deployment without due consideration or consultation with end-users, nor the distros respecting end-users rights to NOT be forced into using it, i cannot and will not be associated or endorse such totally unethical behaviour, so will be removing it from all rootfs images. post-distribution, if people then wish to undo that because they find systemd to be useful and have no objections to its usage they are entirely free to do so.
l.
On Thu, Feb 09, 2017 at 02:59:02PM +0000, Luke Kenneth Casson Leighton wrote:
crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Thu, Feb 9, 2017 at 1:50 PM, Stefan Monnier monnier@iro.umontreal.ca wrote:
W.r.t to logging, I've agree that you're probably better off logging to RAM (or to a remote host) than to a local "disk", and AFAIK that's the default behavior of systemd anyway.
with the exception of fedora which has only a few backers i will NOT be distributing a filesystem which contains the completely unethically-developed and very dangerous systemd application. having evaluated its development, watched the predicted security vulnerabilities unfold and cause massive disruption, and witnessed its "ram it down people's throats" deployment without due consideration or consultation with end-users, nor the distros respecting end-users rights to NOT be forced into using it, i cannot and will not be associated or endorse such totally unethical behaviour, so will be removing it from all rootfs images. post-distribution, if people then wish to undo that because they find systemd to be useful and have no objections to its usage they are entirely free to do so.
I very much like systemd and can hardly see myself using a system without it. Thus I will personally want to have systemd on my systems. Please don't make that too difficult a task for me.
That is: you don't like systemd? fine. Installing Debian without it is rather simple: https://wiki.debian.org/systemd#Installing_without_systemd In that case, all I'll have to do would be to install a few more packages (and maybe disable syslog logging to reduce unnecessary disk writes, and similar tweaks).
However, if "removing it from the rootfs" means something similar to Devuan, then It'll probably be simpler for me to reinstall the image with a proper Debian system.
Please don't make my life complicated.
On Thu, Feb 9, 2017 at 10:37 AM, Tzafrir Cohen tzafrir@cohens.org.il wrote:
On Thu, Feb 09, 2017 at 02:59:02PM +0000, Luke Kenneth Casson Leighton wrote:
crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Thu, Feb 9, 2017 at 1:50 PM, Stefan Monnier monnier@iro.umontreal.ca
wrote:
W.r.t to logging, I've agree that you're probably better off logging to RAM (or to a remote host) than to a local "disk", and AFAIK that's the default behavior of systemd anyway.
with the exception of fedora which has only a few backers i will NOT be distributing a filesystem which contains the completely unethically-developed and very dangerous systemd application. having evaluated its development, watched the predicted security vulnerabilities unfold and cause massive disruption, and witnessed its "ram it down people's throats" deployment without due consideration or consultation with end-users, nor the distros respecting end-users rights to NOT be forced into using it, i cannot and will not be associated or endorse such totally unethical behaviour, so will be removing it from all rootfs images. post-distribution, if people then wish to undo that because they find systemd to be useful and have no objections to its usage they are entirely free to do so.
I very much like systemd and can hardly see myself using a system without it. Thus I will personally want to have systemd on my systems. Please don't make that too difficult a task for me.
That is: you don't like systemd? fine. Installing Debian without it is rather simple: https://wiki.debian.org/systemd#Installing_without_systemd
Please don't spread this mis-information. Installing Debian without systemd is far from simple, and many things just won't work right as systemd has become a dependency on more and more packages.
It's possible, yes, but it's not simple, not supported, and tends to leave the user debugging weird behaviours.
In that case, all I'll have to do would be to install a few more packages (and maybe disable syslog logging to reduce unnecessary disk writes, and similar tweaks).
However, if "removing it from the rootfs" means something similar to Devuan, then It'll probably be simpler for me to reinstall the image with a proper Debian system.
Please don't make my life complicated.
-- Tzafrir Cohen | tzafrir@jabber.org | VIM is http://tzafrir.org.il | | a Mutt's tzafrir@cohens.org.il | | best tzafrir@debian.org | | friend
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
If I may add my own perspective.
Like most people who don't use Win or Mac, I started off my Linux journey with Ubuntu. This was 2007 or so. I eventually (through a story far too long to relate here, particularly since it's not relevant) settled on Puppy Linux of various varieties for my main distro. Puppy, AFAIK, uses BusyBox Init. When SystemD started being developed, it was universally reviled by almost the entire Puppy community... we lost a couple members that way, IIRC.
Anyways, I have heard a lot about SystemD, almost all of it negative. While I have not, to my knowledge, interacted directly with the devs of it, I have heard that they tend to dismiss any bug or issue they are notified with, as "not our problem" -- the underlying attitude being, why make more work for *us* when you can change *your* code to adapt? A similar attitude seems to be prevalent about adopting it -- hey, new kid on the block, let's get to know each other -- and to heck with anyone who doesn't like it.
I find those sorts of attitudes troubling, to say the least. It reminds me a little too well of the Borg, from *Star Trek*...
Yet, my distro of choice right now is not Puppy. (Another painfully long irrelevant story.) I'm using Linux Mint 18.0, and it runs SystemD underneath everything. So far, I have absolutely zero complaints. SystemD has not yet shot my cat, metaphorically or otherwise.
Am I converted? No. I am still skeptical. Would I choose another init system, given the choice? Absolutely. Puppy's cobbled-together nasty pile-of-patches mess of a BusyBox Init served me well for many years and I'd take it back in a heartbeat.
But is SystemD something that keeps me up nights chewing my nails about whether or not my computer will boot tomorrow? Absolutely not.
I have my concerns about how SystemD has been formed and how it propagates. But, by the same token, the people who designed it are not *complete* idiots, regardless of how developed they are (or aren't) socially, because it at least *seems* to work well enough to withstand everyday use... so far at least. I know that SystemD tends to be a divisor amongst Linux users and devs -- you either really love it or you really hate it -- but I find it a little hard to get passionate about it, one way or the other, at this point.
Please don't shoot me. I'm just a messenger :)
On 09/02/2017 15:53, Adam Van Ymeren wrote:
On Thu, Feb 9, 2017 at 10:37 AM, Tzafrir Cohen <tzafrir@cohens.org.il mailto:tzafrir@cohens.org.il> wrote:
On Thu, Feb 09, 2017 at 02:59:02PM +0000, Luke Kenneth Casson Leighton wrote: > --- > crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68 <https://www.crowdsupply.com/eoma68> > > > On Thu, Feb 9, 2017 at 1:50 PM, Stefan Monnier <monnier@iro.umontreal.ca <mailto:monnier@iro.umontreal.ca>> wrote: > > > W.r.t to logging, I've agree that you're probably better off logging to > > RAM (or to a remote host) than to a local "disk", and AFAIK that's the > > default behavior of systemd anyway. > > with the exception of fedora which has only a few backers i will NOT > be distributing a filesystem which contains the completely > unethically-developed and very dangerous systemd application. having > evaluated its development, watched the predicted security > vulnerabilities unfold and cause massive disruption, and witnessed its > "ram it down people's throats" deployment without due consideration or > consultation with end-users, nor the distros respecting end-users > rights to NOT be forced into using it, i cannot and will not be > associated or endorse such totally unethical behaviour, so will be > removing it from all rootfs images. post-distribution, if people then > wish to undo that because they find systemd to be useful and have no > objections to its usage they are entirely free to do so. I very much like systemd and can hardly see myself using a system without it. Thus I will personally want to have systemd on my systems. Please don't make that too difficult a task for me. That is: you don't like systemd? fine. Installing Debian without it is rather simple: https://wiki.debian.org/systemd#Installing_without_systemd <https://wiki.debian.org/systemd#Installing_without_systemd>Please don't spread this mis-information. Installing Debian without systemd is far from simple, and many things just won't work right as systemd has become a dependency on more and more packages.
It's possible, yes, but it's not simple, not supported, and tends to leave the user debugging weird behaviours.
This is not mis-information. Removing systemd from Debian (jessie) is simple, I did only a couple of days ago actually because it screwed with a mail server that was installed on the system.
Who cares if it's supported or not, the Debian devs didn't care whether users wanted it or not in the first place. Can have a link to some examples of 'user debugging weird behaviours'?
What really is simple, is to use Devuan instead :)
Cheers,
Who cares if it's supported or not, the Debian devs didn't care whether users wanted it or not in the first place. Can have a link to some examples of 'user debugging weird behaviours'?
My own experience recently: installing systemd on a Debian testing system fixed the following two bugs I had: - I couldn't control network-manager from nm-applet any more for lack of permissions. - upon wakeup from s2ram, the wifi wouldn't reconnect (it would stay in the state it was before sleeping, thkning it's still connected, but with packets not going anywhere), so I had to kill&restart network-manager every time. Obviously, if you don't use network-manager, you won't be affected.
Stefan
--- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Thu, Feb 9, 2017 at 3:53 PM, Adam Van Ymeren adam.vany@gmail.com wrote:
On Thu, Feb 9, 2017 at 10:37 AM, Tzafrir Cohen tzafrir@cohens.org.il wrote:
On Thu, Feb 09, 2017 at 02:59:02PM +0000, Luke Kenneth Casson Leighton wrote:
crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Thu, Feb 9, 2017 at 1:50 PM, Stefan Monnier monnier@iro.umontreal.ca wrote:
W.r.t to logging, I've agree that you're probably better off logging to RAM (or to a remote host) than to a local "disk", and AFAIK that's the default behavior of systemd anyway.
with the exception of fedora which has only a few backers i will NOT be distributing a filesystem which contains the completely unethically-developed and very dangerous systemd application. having evaluated its development, watched the predicted security vulnerabilities unfold and cause massive disruption, and witnessed its "ram it down people's throats" deployment without due consideration or consultation with end-users, nor the distros respecting end-users rights to NOT be forced into using it, i cannot and will not be associated or endorse such totally unethical behaviour, so will be removing it from all rootfs images. post-distribution, if people then wish to undo that because they find systemd to be useful and have no objections to its usage they are entirely free to do so.
I very much like systemd and can hardly see myself using a system without it. Thus I will personally want to have systemd on my systems. Please don't make that too difficult a task for me.
That is: you don't like systemd? fine. Installing Debian without it is rather simple: https://wiki.debian.org/systemd#Installing_without_systemd
Please don't spread this mis-information. Installing Debian without systemd is far from simple, and many things just won't work right as systemd has become a dependency on more and more packages.
It's possible, yes, but it's not simple, not supported, and tends to leave the user debugging weird behaviours.
for my laptop i'm using angband.pl's nosystemd, which is not kept up-to-date with security updates, so what i have learned is that i can simply "apt-get build-dep" on whatever package is not up-to-date, then "apt-get source" it, build it and install the resultant .debs and that way i can keep running a debian/testing system completely free of systemd and libsystemd0.
... yeah it's not for the faint-hearted :)
l.
On 2017-02-09 at 10:53:02 -0500, Adam Van Ymeren wrote:
On Thu, Feb 9, 2017 at 10:37 AM, Tzafrir Cohen tzafrir@cohens.org.il wrote:
That is: you don't like systemd? fine. Installing Debian without it is rather simple: https://wiki.debian.org/systemd#Installing_without_systemd
Please don't spread this mis-information. Installing Debian without systemd is far from simple, and many things just won't work right as systemd has become a dependency on more and more packages.
It's possible, yes, but it's not simple, not supported, and tends to leave the user debugging weird behaviours.
This is mostly false:
installing debian without libsystemd0 is not supported. Lots of packages have added optional support for systemd, so they are built linking that *648 bytes* library to access it *when available*. Not doing so would require multiple builds of all packages and that would lead to mainteinance hell.
Using another init system on debian is fully supported, mandatory on non-linux archs (sadly, none of them are release archs, but people are still working on them). Since less people are using it it is likely that there are bugs and that it will take more time to find them, but please if you do *report them*, they will be taken care of as any other bug (i.e. not always, because sometimes maintainers disappear).
The only way to be sure that other init systems will die completely in Debian is not reporting bugs, because that way the maintainers have no idea that they exists and no chance to fix them.
Running GNOME without systemd is a different beast: I don't know if it has happened already, but sooner or later systemd will be required because of an *upstream* decision. Debian fully supports a number of other Desktop Environment and window managers, some of which (e.g. KDE/Plasma) have a committed to being multi-platform and thus will not for the foreseeable future force the use of systemd.
On 02/10/2017 10:37 AM, Elena ``of Valhalla'' wrote:
Running GNOME without systemd is a different beast: I don't know if it has happened already, but sooner or later systemd will be required because of an *upstream* decision. Debian fully supports a number of other Desktop Environment and window managers, some of which (e.g. KDE/Plasma) have a committed to being multi-platform and thus will not for the foreseeable future force the use of systemd.
Current GNOME needs systemd interfaces which is why the not yet stable GNU distribution GuixSD uses extracted parts of systemd like elogind to provide them so it can run GNOME without running the systemd init system (which it cannot do on GNU Hurd). So GNOME only needs parts of systemd. See:
https://www.gnu.org/software/guix/news/gnome-in-guixsd.html
I don’t know what Debian GNU/Hurd does. Either way, it’s the distributions’ problem.
Regards, Florian
--- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Thu, Feb 9, 2017 at 3:37 PM, Tzafrir Cohen tzafrir@cohens.org.il wrote:
On Thu, Feb 09, 2017 at 02:59:02PM +0000, Luke Kenneth Casson Leighton wrote:
crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Thu, Feb 9, 2017 at 1:50 PM, Stefan Monnier monnier@iro.umontreal.ca wrote:
W.r.t to logging, I've agree that you're probably better off logging to RAM (or to a remote host) than to a local "disk", and AFAIK that's the default behavior of systemd anyway.
with the exception of fedora which has only a few backers i will NOT be distributing a filesystem which contains the completely unethically-developed and very dangerous systemd application. having evaluated its development, watched the predicted security vulnerabilities unfold and cause massive disruption, and witnessed its "ram it down people's throats" deployment without due consideration or consultation with end-users, nor the distros respecting end-users rights to NOT be forced into using it, i cannot and will not be associated or endorse such totally unethical behaviour, so will be removing it from all rootfs images. post-distribution, if people then wish to undo that because they find systemd to be useful and have no objections to its usage they are entirely free to do so.
I very much like systemd and can hardly see myself using a system without it. Thus I will personally want to have systemd on my systems. Please don't make that too difficult a task for me.
That is: you don't like systemd? fine. Installing Debian without it is rather simple: https://wiki.debian.org/systemd#Installing_without_systemd
does it include removal of libsystemd0? (it doesn't). it's not as straightforward as it's made out to be, tzafrir.
In that case, all I'll have to do would be to install a few more packages (and maybe disable syslog logging to reduce unnecessary disk writes, and similar tweaks).
i'm considering one of two options:
(1) providing the image (a snapshot of debian/testing from before jessie) i've been working with for a couple of years, now, as-is. if people want to upgrade, they just do "apt-get dist-upgrade" and they get systemd and everything else.
(2) putting on angband.pl's nosystemd repositories. this is "hard work" for me to both set up, and for others to remove (revert) just as you say, so i am unlikely to do it... but it's an option.
However, if "removing it from the rootfs" means something similar to Devuan, then It'll probably be simpler for me to reinstall the image with a proper Debian system.
that's why i made devuan available, separately. which, after liking it for a long time i also have issues with: their mission statement says "all-inclusive PID1 choices"... yet systemd is *excluded* from that list. that's disintegritous and so i will not be using devuan. if they gave people the *choice* i'd celebrate and be recommending devuan everywhere and to everyone.
l.
On 09/02/2017 16:45, Luke Kenneth Casson Leighton wrote:
crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Thu, Feb 9, 2017 at 3:37 PM, Tzafrir Cohen tzafrir@cohens.org.il wrote:
On Thu, Feb 09, 2017 at 02:59:02PM +0000, Luke Kenneth Casson Leighton wrote:
crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Thu, Feb 9, 2017 at 1:50 PM, Stefan Monnier monnier@iro.umontreal.ca wrote:
W.r.t to logging, I've agree that you're probably better off logging to RAM (or to a remote host) than to a local "disk", and AFAIK that's the default behavior of systemd anyway.
with the exception of fedora which has only a few backers i will NOT be distributing a filesystem which contains the completely unethically-developed and very dangerous systemd application. having evaluated its development, watched the predicted security vulnerabilities unfold and cause massive disruption, and witnessed its "ram it down people's throats" deployment without due consideration or consultation with end-users, nor the distros respecting end-users rights to NOT be forced into using it, i cannot and will not be associated or endorse such totally unethical behaviour, so will be removing it from all rootfs images. post-distribution, if people then wish to undo that because they find systemd to be useful and have no objections to its usage they are entirely free to do so.
I very much like systemd and can hardly see myself using a system without it. Thus I will personally want to have systemd on my systems. Please don't make that too difficult a task for me.
That is: you don't like systemd? fine. Installing Debian without it is rather simple: https://wiki.debian.org/systemd#Installing_without_systemd
does it include removal of libsystemd0? (it doesn't). it's not as straightforward as it's made out to be, tzafrir.
In that case, all I'll have to do would be to install a few more packages (and maybe disable syslog logging to reduce unnecessary disk writes, and similar tweaks).
i'm considering one of two options:
(1) providing the image (a snapshot of debian/testing from before jessie) i've been working with for a couple of years, now, as-is. if people want to upgrade, they just do "apt-get dist-upgrade" and they get systemd and everything else.
(2) putting on angband.pl's nosystemd repositories. this is "hard work" for me to both set up, and for others to remove (revert) just as you say, so i am unlikely to do it... but it's an option.
However, if "removing it from the rootfs" means something similar to Devuan, then It'll probably be simpler for me to reinstall the image with a proper Debian system.
that's why i made devuan available, separately. which, after liking it for a long time i also have issues with: their mission statement says "all-inclusive PID1 choices"... yet systemd is *excluded* from that list. that's disintegritous and so i will not be using devuan. if they gave people the *choice* i'd celebrate and be recommending devuan everywhere and to everyone.
That's a fair point, but the point of Devuan is/was to provide choice, a choice the Debian devs would not provide, and that is Debian without systemd. So, if you want systemd, go debian, if not, go devuan. Devuan _with_ systemd would be a debian mirror wouldn't it?
Anyway, the main thing is _choice_ and I choose devuan, others can choose whatever, whenever :)
Cheers,
On 02/09/2017 11:45 AM, Luke Kenneth Casson Leighton wrote:
(2) putting on angband.pl's nosystemd repositories. this is "hard work" for me to both set up, and for others to remove (revert) just as you say, so i am unlikely to do it... but it's an option.
for my laptop i'm using angband.pl's nosystemd, which is not kept up-to-date with security updates
Luke, for the love of God and for your own sake, please do not distribute something to your customers that doesn't receive reliable security updates because of your personal preferences. If someone's machine gets compromised because of this, it will tarnish your reputation. Remember your target audience: they're not the type of people to go into a command-line and do...
simply "apt-get build-dep" on whatever package is not up-to-date, then "apt-get source" it, build it and install the resultant .debs
that.
You're advertising computer cards with Parabola, Debian, Fedora, and Devuan pre-installed. You should not attempt to do "more" than that by distributing something that won't update properly (that "testing" image you mentioned) or that might not receive security updates. Just distribute the systems as they are officially distributed, with the exception of the custom kernel needed for the computer card (to ensure that the hardware works correctly) and some custom default settings to disable logging (to avoid needless degradation of the NAND or SD card).
At the risk of repeating myself: Devuan should satisfy the needs of everyone who doesn't like systemd, and I guarantee you that no one who ordered a computer card with Debian wants a version of Debian without systemd.
--- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Thu, Feb 9, 2017 at 5:11 PM, Julie Marchant onpon4@riseup.net wrote:
On 02/09/2017 11:45 AM, Luke Kenneth Casson Leighton wrote:
(2) putting on angband.pl's nosystemd repositories. this is "hard work" for me to both set up, and for others to remove (revert) just as you say, so i am unlikely to do it... but it's an option.
for my laptop i'm using angband.pl's nosystemd, which is not kept up-to-date with security updates
Luke, for the love of God and for your own sake, please do not distribute something to your customers that doesn't receive reliable security updates
i will not be distributing something that doesn't receive reliable security updates. i appreciate you raising this, it's important.
ah. i see where you got the misimpression from. *angband.pl* does not keep up to date with security updates. however that is my personal laptop. i would not recommend *to anyone* that they follow (qty 1000 or 10,000) what i use for my *personal* laptop. it's a highly complex and comprehensive specialist setup.
because of your personal preferences. If someone's machine gets compromised because of this, it will tarnish your reputation. Remember your target audience: they're not the type of people to go into a command-line and do...
then what i might do instead is get debian-installer set up instead. i cannot and will not give something to people that i know will cause them pain or distress, even if they don't understand it.
lots of possibilities, and some time to think about them.
bob's definition of "an ethical act" applies here, so if anyone else can think of something that we've missed, so that it becomes possible (as best we can) to come up with a solution that people are in *100%* agreement is acceptable, that would be... good.
l.
thanks for raising this issue Luke
On 9 February 2017 at 17:21, Luke Kenneth Casson Leighton lkcl@lkcl.net wrote:
crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Thu, Feb 9, 2017 at 5:11 PM, Julie Marchant onpon4@riseup.net wrote:
On 02/09/2017 11:45 AM, Luke Kenneth Casson Leighton wrote:
(2) putting on angband.pl's nosystemd repositories. this is "hard work" for me to both set up, and for others to remove (revert) just as you say, so i am unlikely to do it... but it's an option.
for my laptop i'm using angband.pl's nosystemd, which is not kept up-to-date with security updates
Luke, for the love of God and for your own sake, please do not distribute something to your customers that doesn't receive reliable security updates
i will not be distributing something that doesn't receive reliable security updates. i appreciate you raising this, it's important.
ah. i see where you got the misimpression from. *angband.pl* does not keep up to date with security updates. however that is my personal laptop. i would not recommend *to anyone* that they follow (qty 1000 or 10,000) what i use for my *personal* laptop. it's a highly complex and comprehensive specialist setup.
because of your personal preferences. If someone's machine gets compromised because of this, it will tarnish your reputation. Remember your target audience: they're not the type of people to go into a command-line and do...
then what i might do instead is get debian-installer set up instead. i cannot and will not give something to people that i know will cause them pain or distress, even if they don't understand it.
lots of possibilities, and some time to think about them.
bob's definition of "an ethical act" applies here, so if anyone else can think of something that we've missed, so that it becomes possible (as best we can) to come up with a solution that people are in *100%* agreement is acceptable, that would be... good.
l.
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
It got really heated in here Dear Luke, please just ship stock versions of the distros you promised. Customers will feel very cheated if they don't get what you promised. If SystemD nukes the card as it were, then it'll be Debian devs problem not yours, as you already provide libre well documented hardware. If indeed you want systemd-less versions of the distros you promised let them be officially supported ones, but I repeat yet again, don't ship a Frankendistro.
On February 9, 2017 8:21:56 PM GMT+03:00, Luke Kenneth Casson Leighton lkcl@lkcl.net wrote:
crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Thu, Feb 9, 2017 at 5:11 PM, Julie Marchant onpon4@riseup.net wrote:
On 02/09/2017 11:45 AM, Luke Kenneth Casson Leighton wrote:
(2) putting on angband.pl's nosystemd repositories. this is "hard work" for me to both set up, and for others to remove (revert) just
as
you say, so i am unlikely to do it... but it's an option.
for my laptop i'm using angband.pl's nosystemd, which is not kept up-to-date with security updates
Luke, for the love of God and for your own sake, please do not distribute something to your customers that doesn't receive reliable security updates
i will not be distributing something that doesn't receive reliable security updates. i appreciate you raising this, it's important.
ah. i see where you got the misimpression from. *angband.pl* does not keep up to date with security updates. however that is my personal laptop. i would not recommend *to anyone* that they follow (qty 1000 or 10,000) what i use for my *personal* laptop. it's a highly complex and comprehensive specialist setup.
because of your personal preferences. If someone's machine gets compromised because of this, it will tarnish your reputation. Remember your target audience: they're not the type of people to go into a command-line and do...
then what i might do instead is get debian-installer set up instead. i cannot and will not give something to people that i know will cause them pain or distress, even if they don't understand it.
lots of possibilities, and some time to think about them.
bob's definition of "an ethical act" applies here, so if anyone else can think of something that we've missed, so that it becomes possible (as best we can) to come up with a solution that people are in *100%* agreement is acceptable, that would be... good.
l.
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
On Thu, Feb 09, 2017 at 04:45:07PM +0000, Luke Kenneth Casson Leighton wrote:
On Thu, Feb 9, 2017 at 3:37 PM, Tzafrir Cohen tzafrir@cohens.org.il wrote:
That is: you don't like systemd? fine. Installing Debian without it is rather simple: https://wiki.debian.org/systemd#Installing_without_systemd
does it include removal of libsystemd0? (it doesn't). it's not as straightforward as it's made out to be, tzafrir.
Neither it include the removal of libselinux. I am yet to hear proper technical arguments why this library should be removed.
In that case, all I'll have to do would be to install a few more packages (and maybe disable syslog logging to reduce unnecessary disk writes, and similar tweaks).
i'm considering one of two options:
(1) providing the image (a snapshot of debian/testing from before jessie) i've been working with for a couple of years, now, as-is. if people want to upgrade, they just do "apt-get dist-upgrade" and they get systemd and everything else.
I would rather start with a fresh install of Debian than this.
(2) putting on angband.pl's nosystemd repositories. this is "hard work" for me to both set up, and for others to remove (revert) just as you say, so i am unlikely to do it... but it's an option.
I would rather start with a fresh install of Debian than this.
Neither of those options is a properly supported Debian system. This is not what I meant when I opted for Debian as the OS (IIRC this was also the most common option among others). I guess some of us will need to provide an alternative image of Debian with instructions on how to install it.
I agree with Julie here. If you tell users you're providing Debian, it should be stock Debian (or as close as possible as is needed to support the hardware). Likewise for every other distro.
I would be incredibly irritated to get a "Debian" computer card, only to find out that some of the software I planned to use doesn't work properly because the distributor of the card decided to tweak the software before sending it to me. With a stock Debian install, it's Debian's responsibility if something doesn't work properly, not yours.
On Thursday 9. February 2017 21.23.19 Jonathan Frederickson wrote:
I agree with Julie here. If you tell users you're providing Debian, it should be stock Debian (or as close as possible as is needed to support the hardware). Likewise for every other distro.
I agree with Jonathan and Julie here. ;-)
Let the distinction between Debian and Devuan be the means by which people choose to run systemd or not. After all, that is more or less the reason why there are two such distributions rather than one. [*]
By all means indicate why you do not agree with systemd, but instead of making more work for yourself, may I respectfully suggest that you just let people switch their order from Debian to Devuan if they agree with you now, perhaps not having thought about the matter before?
And let people switch the other way (or to something else) if they agree with your view of Devuan and actually want the option of having systemd from first power-on. As for people who agree with you on both systemd and Devuan's attitude towards it, I guess you could just see if this affects anyone who placed an order.
Paul
[*] There are supposedly improvements in the distribution-building process with Devuan, which I thought could be of interest to various libre distributions - removing unwanted packages and content is what they do, after all - but the documentation isn't that great for any of these distributions in informing what the possibilities might be.
Tzafrir Cohen tzafrir@cohens.org.il writes:
On Thu, Feb 09, 2017 at 04:45:07PM +0000, Luke Kenneth Casson Leighton wrote:
On Thu, Feb 9, 2017 at 3:37 PM, Tzafrir Cohen tzafrir@cohens.org.il wrote:
That is: you don't like systemd? fine. Installing Debian without it is rather simple: https://wiki.debian.org/systemd#Installing_without_systemd
does it include removal of libsystemd0? (it doesn't). it's not as straightforward as it's made out to be, tzafrir.
Neither it include the removal of libselinux. I am yet to hear proper technical arguments why this library should be removed.
In that case, all I'll have to do would be to install a few more packages (and maybe disable syslog logging to reduce unnecessary disk writes, and similar tweaks).
i'm considering one of two options:
(1) providing the image (a snapshot of debian/testing from before jessie) i've been working with for a couple of years, now, as-is. if people want to upgrade, they just do "apt-get dist-upgrade" and they get systemd and everything else.
I would rather start with a fresh install of Debian than this.
(2) putting on angband.pl's nosystemd repositories. this is "hard work" for me to both set up, and for others to remove (revert) just as you say, so i am unlikely to do it... but it's an option.
I would rather start with a fresh install of Debian than this.
Neither of those options is a properly supported Debian system. This is not what I meant when I opted for Debian as the OS (IIRC this was also the most common option among others). I guess some of us will need to provide an alternative image of Debian with instructions on how to install it.
Quite.
Luke, you're going to have some seriously unhappy customers if you start trying to supply mangled versions of Debian in place of the advertised "Debian GNU/Linux".
I'm sure that Tzafrir and I are not the only Debian Developers who have paid you, and knew _exactly_ what to expect from Debian when they did so.
It strikes me as seriously condescending to assume that Debian users are somehow clueless about what they mean by Debian GNU/Linux. Especially when offered the choice of OSs you presented.
Deciding to supply something else, now, strikes me as ... impolite.
If you are concerned about ethics, consider the ethics of bait-and-switch, which is what you appear to be contemplating here.
Cheers, Phil.
Yes, I would be an unhappy customer, I ordered the original debian and nothing else.
And BTW Luke, if you do not want to deliver unethical products, you can not deliver hardware. There is not such a thing as ethical harware: conflict minerals, working conditions, poisoning the enviroment, ...
We should be glad, if we get open and free hardware in the near future with running Free software on top. But that would be not ehtical produced hardware. It is a very, very long way to go, until we get that.
Wolfgang
Am 09.02.2017 um 22:56 schrieb Philip Hands:
Tzafrir Cohen tzafrir@cohens.org.il writes:
On Thu, Feb 09, 2017 at 04:45:07PM +0000, Luke Kenneth Casson Leighton wrote:
On Thu, Feb 9, 2017 at 3:37 PM, Tzafrir Cohen tzafrir@cohens.org.il wrote:
That is: you don't like systemd? fine. Installing Debian without it is rather simple: https://wiki.debian.org/systemd#Installing_without_systemd
does it include removal of libsystemd0? (it doesn't). it's not as straightforward as it's made out to be, tzafrir.
Neither it include the removal of libselinux. I am yet to hear proper technical arguments why this library should be removed.
In that case, all I'll have to do would be to install a few more packages (and maybe disable syslog logging to reduce unnecessary disk writes, and similar tweaks).
i'm considering one of two options:
(1) providing the image (a snapshot of debian/testing from before jessie) i've been working with for a couple of years, now, as-is. if people want to upgrade, they just do "apt-get dist-upgrade" and they get systemd and everything else.
I would rather start with a fresh install of Debian than this.
(2) putting on angband.pl's nosystemd repositories. this is "hard work" for me to both set up, and for others to remove (revert) just as you say, so i am unlikely to do it... but it's an option.
I would rather start with a fresh install of Debian than this.
Neither of those options is a properly supported Debian system. This is not what I meant when I opted for Debian as the OS (IIRC this was also the most common option among others). I guess some of us will need to provide an alternative image of Debian with instructions on how to install it.
Quite.
Luke, you're going to have some seriously unhappy customers if you start trying to supply mangled versions of Debian in place of the advertised "Debian GNU/Linux".
I'm sure that Tzafrir and I are not the only Debian Developers who have paid you, and knew _exactly_ what to expect from Debian when they did so.
It strikes me as seriously condescending to assume that Debian users are somehow clueless about what they mean by Debian GNU/Linux. Especially when offered the choice of OSs you presented.
Deciding to supply something else, now, strikes me as ... impolite.
If you are concerned about ethics, consider the ethics of bait-and-switch, which is what you appear to be contemplating here.
Cheers, Phil.
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
On 02/09/2017 09:59 AM, Luke Kenneth Casson Leighton wrote:
with the exception of fedora which has only a few backers i will NOT be distributing a filesystem which contains the completely unethically-developed and very dangerous systemd application. having evaluated its development, watched the predicted security vulnerabilities unfold and cause massive disruption, and witnessed its "ram it down people's throats" deployment without due consideration or consultation with end-users, nor the distros respecting end-users rights to NOT be forced into using it, i cannot and will not be associated or endorse such totally unethical behaviour, so will be removing it from all rootfs images. post-distribution, if people then wish to undo that because they find systemd to be useful and have no objections to its usage they are entirely free to do so.
I don't personally care whether or not my system includes systemd (it does), but I have seen no evidence of wrongdoing by the systemd developers. Every single case of systemd adoption I am aware of was because the maintainers of a distro wanted to adopt it. And systemd is libre, even copylefted. There are no ethical grounds to oppose it.
Besides, people who hate systemd would choose Devuan rather than Debian. Why would you give people who ordered a computer card with Debian (i.e. using systemd) something they didn't ask for? Best-case scenario, no one cares. Worst-case scenario, you make a decision that turns out to be a maintenance burden for you and annoys some of your backers. Regardless of your opinion on the matter, this isn't a battle you should be fighting, at least not here.
On Thu, Feb 9, 2017 at 3:49 PM, Julie Marchant onpon4@riseup.net wrote:
I don't personally care whether or not my system includes systemd (it does), but I have seen no evidence of wrongdoing by the systemd developers. Every single case of systemd adoption I am aware of was because the maintainers of a distro wanted to adopt it. And systemd is libre, even copylefted. There are no ethical grounds to oppose it.
you're right... at face value. it's complicated, and took me a long time to work out what the problem is, and even longer to be able to express it.
Besides, people who hate systemd would choose Devuan rather than Debian. Why would you give people who ordered a computer card with Debian (i.e. using systemd) something they didn't ask for? Best-case scenario, no one cares. Worst-case scenario, you make a decision that turns out to be a maintenance burden for you and annoys some of your backers. Regardless of your opinion on the matter, this isn't a battle you should be fighting, at least not here.
i have some rules, julie, which are not really negotiable. i cannot provide something to people which i *know* will cause them pain and anguish. it's simply not possible for me to do that, even if they themselves are not able to follow the (potentially complex) logic which led me to conclude that, by providing something to them, it would cause them distress at some point in their future.
the *whole project* is based on that premise. if i start deliberately and consciously compromising *even once* then i am screwed and cannot be trusted publicly to ever honor my commitments to integrity of purpose, ever again. mistakes, yeah fine. *deliberate* compromises that cost me my integrity: i'm done, and the entire project's a failure.
so the consequences are much more severe than it seems at face value.
you noticed that all the individual distros made a choice. they made those decisions freely and without consultation with other distros because their job is to focus on *their* distro.
and that the systemd team, working in isolation from distros and only having links to the software package writers, also made their technical decisions as best they can.
then, also, the package writers, working in isolation, because it's not their job to be a distro maintainer or an init PID1 software developer, *also* made their decisions freely and perfectly well, without wider consultation, because that's *also* their job.
so it's important to note: there is *nothing wrong* with these separate processes, *nobody did anything wrong*. each team is *perfectly* executing their localised strategy for the development and maintenance of *their* project.
taken collectively, however, is where the problems start. the *ENTIRE* free software community, like a "shoal of fish", suddenly switched direction, without warning, without thinking, and, crucially, WITHOUT CONSULTING THE USERS.
now, if we take for example debian, we know that it has a charter. it's a written contract that everyone understands and accepts. the bits that are *NOT* included - *NOT* written down - but are "unspoken", is that the end-users place their trust in the debian maintainers to not disrupt their lives by making decisions that would force them to take drastic and costly action.
this UNSPOKEN and UNWRITTEN agreement is what the debian developers very unfortunately violated by ignoring the vote on default choice of init system which SPECIFICALLY concluded that systemd would be the absolute worst possible choice to make. they paid the price for that decision with the loss of many key strategic developers, and are beginning to make amends by incorporating several init systems into the current debian/testing (but still leaving libsystemd0 in place).
archlinux, as a smaller community which is a running distro, fared much better, because it is a tight-knit community where all its users are required to keep much more up-to-date. thus there was much less of a problem. as they are a little bit more technically-minded, several of them created community-based repos that allow systemd to be replaced entirely by sysvinit, and udev to be replaced with eudev. it works very well and is faster to boot on the a20. it's also seamless and completely non-disruptive. the faster turnaround time on archlinux allowed debugging and testing to be completed rapidly.
now, from a technical perspective, the *specific* technical issues with systemd are best expressed by andrew tridgell's evaluation of systemd, which can be found on the samba mailing lists. the warnings that he outlined would come true *have* come true - several times already, in the form of the security vulnerabilities that he predicted would occur. remember: andrew is an extremely experienced low-level systems programmer and reverse-engineer, who has had to deal with some extreme attacks against samba, so he's very knowledgeable about how to code and design software in a strategic fashion that will mitigate against attacks.
summary: he's *deeply* unimpressed with the design and continued escalating scope-creep of systemd, but, unlike many people who warn or complain about it, he actually *knows* what he's talking about.
so do i... but i lack his ability to vocalise my thoughts in a short-term timeframe, it tends to take me much longer to be able to express things clearly (by which time often it's far too late *sigh*).
anyway, bottom line: from what i know of systemd (that includes libsystemd0) i know the pain that it will bring people, and because of that i cannot possibly distribute it to others. it would be a fundamental violation of my ethical principles to do so. i'm *really* not happy with the fact that i'll be supplying it to the backers who pledged for fedora cards. i'll have to think how to deal with that.
l.
On 02/09/2017 12:15 PM, Luke Kenneth Casson Leighton wrote:
i have some rules, julie, which are not really negotiable. i cannot provide something to people which i *know* will cause them pain and anguish. it's simply not possible for me to do that, even if they themselves are not able to follow the (potentially complex) logic which led me to conclude that, by providing something to them, it would cause them distress at some point in their future.
I've been using systemd ever since I installed Debian 8, and then Ubuntu 16.04. It's the same experience: I push the power button, and it boots successfully every time. It hasn't caused me "pain" or "distress".
What *could* cause me distress is if I ordered a computer with Debian on it, and I instead got some modified version that doesn't work the same way. Another thing that could cause me distress is if a problem occurs that just doesn't happen in stock Debian because of unnecessary modifications you made. Anyone experiencing distress for either of these reasons is not going to be persuaded by your vague "I looked into it, systemd is terrible" stance. They will simply stop trusting you.
the *whole project* is based on that premise. if i start deliberately and consciously compromising *even once* then i am screwed and cannot be trusted publicly to ever honor my commitments to integrity of purpose, ever again. mistakes, yeah fine. *deliberate* compromises that cost me my integrity: i'm done, and the entire project's a failure.
Where in the EOMA68 campaign did you say that you would purge systemd from the Debian and Parabola systems you distribute?
For the record, if you had said that, I would have only ordered a computer card with the understanding that I would have to wipe clean whatever system is pre-installed on it and install my system of choice myself. Not because I want systemd per se, but because I've used an unofficial barely maintained spin of Debian before (Pandian), and it was not a great experience.
As for not compromising, the only compromises you should avoid are *ruinous* compromises. Is offering systems that have systemd on them a ruinous compromise? I don't think so. The only issues raised by it are technical issues. If there is any reason that systemd is *unethical*, no one to my knowledge has raised that as an issue.
If your principles include refusing to give something to someone simply because of *technical* inferiority, that's just a bad principle. You seem to be an intelligent guy, but you don't know everything, and in particular you are not in a position to judge the entire world's technical needs. If you don't think Debian is something that fits people's needs, it's fine to not offer it, but if you do agree to put Debian on the card, that should be *stock Debian* to the best of your ability, not some custom version of it.
you noticed that all the individual distros made a choice. they made those decisions freely and without consultation with other distros because their job is to focus on *their* distro.
and that the systemd team, working in isolation from distros and only having links to the software package writers, also made their technical decisions as best they can.
then, also, the package writers, working in isolation, because it's not their job to be a distro maintainer or an init PID1 software developer, *also* made their decisions freely and perfectly well, without wider consultation, because that's *also* their job.
so it's important to note: there is *nothing wrong* with these separate processes, *nobody did anything wrong*. each team is *perfectly* executing their localised strategy for the development and maintenance of *their* project.
taken collectively, however, is where the problems start. the *ENTIRE* free software community, like a "shoal of fish", suddenly switched direction, without warning, without thinking, and, crucially, WITHOUT CONSULTING THE USERS.
Luke, that's standard practice in libre software development. No one voted on GNOME 3, the recent interface change for Firefox, or Linux as the kernel everyone uses. These all just *happened*. Some are the result of individual projects, and in that case you get forks like MATE and Pale Moon. Some are a result of implied consensus, like the adoption of Linux and now systemd.
But in any case, no one is ever forced to go with the changes. They go along with the changes either because they agree with them, or they don't care. Now, the people who don't care can cause social inertia to favor things that are bad, like proprietary programs, but you have to demonstrate that those things are bad. You can't just say that *everything* that's allowed to happen because the users don't care is a problem, and if there is a problem you can identify, the fact that it's being adopted because of social inertia from people who don't care is unimportant.
And like you admitted here, the systemd developers did nothing wrong. So how, then, is it appropriate to boycott systemd?
now, if we take for example debian, we know that it has a charter. it's a written contract that everyone understands and accepts. the bits that are *NOT* included - *NOT* written down - but are "unspoken", is that the end-users place their trust in the debian maintainers to not disrupt their lives by making decisions that would force them to take drastic and costly action.
this UNSPOKEN and UNWRITTEN agreement is what the debian developers very unfortunately violated by ignoring the vote on default choice of init system which SPECIFICALLY concluded that systemd would be the absolute worst possible choice to make. they paid the price for that decision with the loss of many key strategic developers, and are beginning to make amends by incorporating several init systems into the current debian/testing (but still leaving libsystemd0 in place).
First of all, Debian made a change for its *next* release, i.e. during the testing period. There is not an unwritten agreement that the testing release will be stable; only the stable releases are stable. People who are using Debian still have time *now* before they need to upgrade to the version of Debian that uses systemd, and heck, there's nothing stopping them from continuing to use sysvinit if they really want to. I find it baffling that you consider this to be something that "disrupts their lives", and not the Iceweasel updates which always come in due to the way Firefox is developed. Because of security concerns, Debian Stable actually adds major updates to Iceweasel. Yeah, it sticks with ESR releases when possible, but when the next ESR release comes in, it's updated to that. So you have the potential (although unlikely) for browser add-on breakage *within* a stable release, and what you're worried about is the potential (although unlikely) for init script breakage *between* releases?
Secondly, it just isn't true that systemd was voted as the "absolute worst possible choice to make". systemd was tied for *first place* with Upstart.[1] In fact, *sysvinit* was voted as least favorable.
But ultimately, if you think that *Debian* did something wrong, why are you pinning this on systemd? If you're going to take a hardline stance based on the way that Debian switched to systemd, then surely you should be refusing to give people Debian, period.
[1] https://lists.debian.org/debian-ctte/2014/02/msg00402.html
Le jeu. 9 févr. 2017 à 13:38, Julie Marchant onpon4@riseup.net a écrit :
Where in the EOMA68 campaign did you say that you would purge systemd from the Debian and Parabola systems you distribute?
parabola openrc is also a thing, i'm neither using nor maintaining it, but others are, and updates/support should not be worse than systemd's.
On 02/09/2017 12:15 PM, Luke Kenneth Casson Leighton wrote:
On Thu, Feb 9, 2017 at 3:49 PM, Julie Marchant onpon4@riseup.net wrote:
I don't personally care whether or not my system includes systemd (it does), but I have seen no evidence of wrongdoing by the systemd developers. Every single case of systemd adoption I am aware of was because the maintainers of a distro wanted to adopt it. And systemd is libre, even copylefted. There are no ethical grounds to oppose it.
you're right... at face value. it's complicated, and took me a long time to work out what the problem is, and even longer to be able to express it.
Besides, people who hate systemd would choose Devuan rather than Debian. Why would you give people who ordered a computer card with Debian (i.e. using systemd) something they didn't ask for? Best-case scenario, no one cares. Worst-case scenario, you make a decision that turns out to be a maintenance burden for you and annoys some of your backers. Regardless of your opinion on the matter, this isn't a battle you should be fighting, at least not here.
i have some rules, julie, which are not really negotiable. i cannot provide something to people which i *know* will cause them pain and anguish. it's simply not possible for me to do that, even if they themselves are not able to follow the (potentially complex) logic which led me to conclude that, by providing something to them, it would cause them distress at some point in their future.
the *whole project* is based on that premise. if i start deliberately and consciously compromising *even once* then i am screwed and cannot be trusted publicly to ever honor my commitments to integrity of purpose, ever again. mistakes, yeah fine. *deliberate* compromises that cost me my integrity: i'm done, and the entire project's a failure.
so the consequences are much more severe than it seems at face value.
you noticed that all the individual distros made a choice. they made those decisions freely and without consultation with other distros because their job is to focus on *their* distro.
and that the systemd team, working in isolation from distros and only having links to the software package writers, also made their technical decisions as best they can.
then, also, the package writers, working in isolation, because it's not their job to be a distro maintainer or an init PID1 software developer, *also* made their decisions freely and perfectly well, without wider consultation, because that's *also* their job.
so it's important to note: there is *nothing wrong* with these separate processes, *nobody did anything wrong*. each team is *perfectly* executing their localised strategy for the development and maintenance of *their* project.
taken collectively, however, is where the problems start. the *ENTIRE* free software community, like a "shoal of fish", suddenly switched direction, without warning, without thinking, and, crucially, WITHOUT CONSULTING THE USERS.
now, if we take for example debian, we know that it has a charter. it's a written contract that everyone understands and accepts. the bits that are *NOT* included - *NOT* written down - but are "unspoken", is that the end-users place their trust in the debian maintainers to not disrupt their lives by making decisions that would force them to take drastic and costly action.
this UNSPOKEN and UNWRITTEN agreement is what the debian developers very unfortunately violated by ignoring the vote on default choice of init system which SPECIFICALLY concluded that systemd would be the absolute worst possible choice to make. they paid the price for that decision with the loss of many key strategic developers, and are beginning to make amends by incorporating several init systems into the current debian/testing (but still leaving libsystemd0 in place). archlinux, as a smaller community which is a running distro, fared much better, because it is a tight-knit community where all its users are required to keep much more up-to-date. thus there was much less of a problem. as they are a little bit more technically-minded, several of them created community-based repos that allow systemd to be replaced entirely by sysvinit, and udev to be replaced with eudev. it works very well and is faster to boot on the a20. it's also seamless and completely non-disruptive. the faster turnaround time on archlinux allowed debugging and testing to be completed rapidly.
now, from a technical perspective, the *specific* technical issues with systemd are best expressed by andrew tridgell's evaluation of systemd, which can be found on the samba mailing lists. the warnings that he outlined would come true *have* come true - several times already, in the form of the security vulnerabilities that he predicted would occur. remember: andrew is an extremely experienced low-level systems programmer and reverse-engineer, who has had to deal with some extreme attacks against samba, so he's very knowledgeable about how to code and design software in a strategic fashion that will mitigate against attacks.
S does Systemd have security issues? privacy issues? stability issues?
or a combination of two or all three?
If you say yes to one or more,
then probably people shouldn't be using systemd... at least if it has more issues than the other options anyways.
which I am assuming is the case.
right Luke?
I may not know all the issues so please enlighten me further.
summary: he's *deeply* unimpressed with the design and continued escalating scope-creep of systemd, but, unlike many people who warn or complain about it, he actually *knows* what he's talking about.
so do i... but i lack his ability to vocalise my thoughts in a short-term timeframe, it tends to take me much longer to be able to express things clearly (by which time often it's far too late *sigh*).
anyway, bottom line: from what i know of systemd (that includes libsystemd0) i know the pain that it will bring people, and because of that i cannot possibly distribute it to others. it would be a fundamental violation of my ethical principles to do so. i'm *really* not happy with the fact that i'll be supplying it to the backers who pledged for fedora cards. i'll have to think how to deal with that.
l.
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
On 02/09/2017 06:11 PM, zap wrote:
security issues? privacy issues? stability issues?
No on all three counts. The only thing "wrong" with systemd is that some people don't like it.
Kind of like how I can't stand vi. But that's a whole other discussion. :P
Well a quick web search (though I couldn't quite find the reference to the mailing list reference from Andrew at Samba) shows that the issue is that it is a) a change from unix style services that do something well and don't overtake the rest of the system b) a system that increasingly does more (whether that's negotiate DNS for you and/or log in a binary format) and more besides. It's true that I've no idea if any of those sources on the internet are super trustworthy. But I did already kill my distro's setup a little, so I can kinda see how sudo/su/and a host of other services can accidentally on purpose die just because of the init system changing (like if you (well, I personally in this instance) modify or dpkg delete (whatever the vernacular for that is) systemd (or partially))
As to my order, not that it probably matters, as I ordered the LibreTea card. (I'm not sure if that'll come with systemd since libsystemd is (perhaps) free in some senses of the term. But increasingly less, as per Luke's post.
Though, I also see your points about wanting to have expected behaviour, and seeing a distro as whatever that distro is (good or bad).
Happy hacking,
Russell
On 9 February 2017 at 23:23, Julie Marchant onpon4@riseup.net wrote:
On 02/09/2017 06:11 PM, zap wrote:
security issues? privacy issues? stability issues?
No on all three counts. The only thing "wrong" with systemd is that some people don't like it.
Kind of like how I can't stand vi. But that's a whole other discussion. :P
-- Julie Marchant https://onpon4.github.io
Protect your emails with GnuPG: https://emailselfdefense.fsf.org
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
Oops, I left out some web links, here's one I found: http://without-systemd.org/wiki/index.php/Main_Page
On 10 February 2017 at 00:44, Russell Hyer russell.hyer@gmail.com wrote:
Well a quick web search (though I couldn't quite find the reference to the mailing list reference from Andrew at Samba) shows that the issue is that it is a) a change from unix style services that do something well and don't overtake the rest of the system b) a system that increasingly does more (whether that's negotiate DNS for you and/or log in a binary format) and more besides. It's true that I've no idea if any of those sources on the internet are super trustworthy. But I did already kill my distro's setup a little, so I can kinda see how sudo/su/and a host of other services can accidentally on purpose die just because of the init system changing (like if you (well, I personally in this instance) modify or dpkg delete (whatever the vernacular for that is) systemd (or partially))
As to my order, not that it probably matters, as I ordered the LibreTea card. (I'm not sure if that'll come with systemd since libsystemd is (perhaps) free in some senses of the term. But increasingly less, as per Luke's post.
Though, I also see your points about wanting to have expected behaviour, and seeing a distro as whatever that distro is (good or bad).
Happy hacking,
Russell
On 9 February 2017 at 23:23, Julie Marchant onpon4@riseup.net wrote:
On 02/09/2017 06:11 PM, zap wrote:
security issues? privacy issues? stability issues?
No on all three counts. The only thing "wrong" with systemd is that some people don't like it.
Kind of like how I can't stand vi. But that's a whole other discussion. :P
-- Julie Marchant https://onpon4.github.io
Protect your emails with GnuPG: https://emailselfdefense.fsf.org
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
On Fri, Feb 10, 2017 at 12:44:42AM +0000, Russell Hyer wrote:
Oops, I left out some web links, here's one I found: http://without-systemd.org/wiki/index.php/Main_Page
Which has tons of links and no actual content. I figure that most linked pages generally recycle content. So I'm looking at content of the first link from the first link: http://without-systemd.org/wiki/index.php/Local_copy_of_boycottsystemd.org_a...
Answering just the first items:
1. False. It can handle hostname, locale, and such. But doesn't have to. For instance, systemd introduced systemd-localed and localectl. However, you don't like them? just disable systemd-localed.service and use whatever method you used for locale settings.
The only extra component of systemd that is required is the journal.
2. Somewhat true. However, I have hardly had cases of he journal files corrupted. Syslog test files are likewise not ACID compliant, but this is irrelevant when you have a single writer.
3. Somewhat true. Right now systemd requires kernel >= 3.12, IIRC.
4. Somewhat true. Both udev and dbus have been part of most Linux systems (certainly desktop ones) for quite some time. However, the separate dbus transport is not intended to move dbus into systemd. Rather, it intends mostly to move it into the kernel and thus improve efficiency.
5. Misleading. Debian does not use that interface by default. Which means it is an optional component. That said, it is very handy to use.
6. Somewhat true. How about all the CVEs systemd saved by avoiding poorly written scripts (with poorly written temporary files handling and other races)? By making it simple to drop privileges? And other cases? Or using timedatectl insteead of ntpd?
That said, I have no idea what those entries refer to. So I hope those numbers are corect.
I suppose it won't get much better.
--- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Sun, Feb 12, 2017 at 9:10 AM, Tzafrir Cohen tzafrir@cohens.org.il wrote:
- Somewhat true. Right now systemd requires kernel >= 3.12, IIRC.
ah - then the discussion is over, as it is necessary for the Cards to go out with the sunxi 3.4 kernel by default, as this is the only stable (full featured) kernel available.
a stable kernel, clearly, takes top precedence.
once the first Cards are out there, i would assume it would be natural for people to help each other get things up-to-date and to sort out any software issues.
that leaves me free to focus on hardware design (which i would like to also be moved to the responsibility of others) and on being the guardian of the eoma68 standard.
l.
On February 12, 2017 1:40:50 AM PST, Luke Kenneth Casson Leighton lkcl@lkcl.net wrote:
once the first Cards are out there, i would assume it would be natural for people to help each other get things up-to-date and to sort out any software issues.
that leaves me free to focus on hardware design (which i would like to also be moved to the responsibility of others) and on being the guardian of the eoma68 standard.
Indeed, this us vs. l. attitude isn't productive. We are a community built around a standards project. I think l. is hard at work to get a product out the door with the software that is tested and working on it without unnecessary delay, as part of the project. If any of us are dissatisfied with the idea of what's on the product, then one or more of us ought to get cracking on producing some of those "specially prepared microSD card"s that will flash a proper image onto the product, as another part of the project.
-- Eric Duhamel http://www.noxbanners.net/
On Sun, Feb 12, 2017 at 10:20 AM, Eric Duhamel ericxdu23@gmail.com wrote:
Indeed, this us vs. l. attitude isn't productive. We are a community built around a standards project. I think l. is hard at work to get a product out the door with the software that is tested and working on it without unnecessary delay, as part of the project. If any of us are dissatisfied with the idea of what's on the product, then one or more of us ought to get cracking on producing some of those "specially prepared microSD card"s that will flash a proper image onto the product, as another part of the project.
thank you eric. update due soon which asks pretty much exactly this. so, thank you.
l.
On Thu, 09 Feb 2017 17:15:28 +0000, Luke Kenneth Casson Leighton wrote:
anyway, bottom line: from what i know of systemd (that includes libsystemd0) i know the pain that it will bring people, and because of that i cannot possibly distribute it to others. it would be a fundamental violation of my ethical principles to do so. i'm *really* not happy with the fact that i'll be supplying it to the backers who pledged for fedora cards. i'll have to think how to deal with that.
Warn them. And send them fedora if they really want it or ignore the warning. Offer them Devuan if they heed the warnings.
The same for Debian, especially.
Most of them will not heed the warnings, but *if* they run into trouble they will think of the warning and probably not blame you.
And make Devuan an option for those pledging.
-- hendrik
with the exception of fedora which has only a few backers i will NOT be distributing a filesystem which contains the completely unethically-developed and very dangerous systemd application.
I mentioned systemd simply as a data point indicating that its considered normal nowadays to log to ram.
cares. Worst-case scenario, you make a decision that turns out to be a maintenance burden for you and annoys some of your backers. Regardless
FWIW, I recently "solved" some odd permission problems I was having with network-manager as well as a few other minor oddities on a Debian testing system by installing systemd on it (I hadn't worked hard to avoid it, but that 10 year old installation of Debian testing somehow had managed to go through incremental upgrades without install systemd).
So, yes, I wouldn't be surprised that Debian-without-systemd suffers from latent bugs.
Stefan
ok, so, apologies for not responding for 2 days: the latest cold, which is back again a fourth time in as many weeks, is leaving me exhausted. again.
tzafrir: i've mentioned this a number of times, and am happy to mention it again, as you appear to have missed it. the key difference is massive scope-creep. look at how the NSA developed libselinux1 and associated infrastructure. they got a university involved to develop the FLASK model. they set out a design strategy, they set out what they were going to do, then they did it. whilst almost everything else that the NSA does may be questionable, steven smalley is clearly a smart guy and knows what he's doing.
by contrast the development of systemd has become a critical single-point of failure for a massive number of distros, where its developers are clearly and pathologically not taking responsibility for the consequences of their *technically-driven* decisions, and are continuing to develop their software without wider consultation.
so, i read what everyone wrote: i think the simplest thing to do is to just go with the image that i have been working with and testing over the past two years. it's using xfce4 (gnome is too heavy). i know it works, and i simply don't have the time - or importantly the energy - to create a new image, *especially* based on people's comments and reactions that they'd be deeply unhappy with it not being a "stock image", even if all i did was make it boot sysvinit instead by default. those comments *alone* immediately terminate all and any possibility that i can provide debian/jessie in a 100% ethical way.
on receipt of the cards, anybody who wants to will be free and entirely at liberty to do "apt-get dist-upgrade" and they will be taking direct responsibility for doing so. for those people who are technically-minded they are also entirely free and at liberty to set up a from-scratch root filesystem.
this is a decision that is easily justifiable based on the fact that it's going to have to be distributed with the sunxi 3.4 kernel as that's the only one which supports the full hardware.
l.
https://www.theregister.co.uk/2017/01/24/systemd_flaw/
"Newer" versions of systemd deployed by Fedora or Ubuntu have been secured, but Debian systems are still running an older version and therefore need updating.
systemd is a suite for building blocks for Linux systems that provides system and service management technology. Security specialists view it with suspicion and ***>>>complaints about function creep are not uncommon<<<***.
https://betanews.com/2016/10/07/systemd-vulnerability-linux-crash/
The reason he has decided to disclose the bug publicly was to bring further attention to problems with a widely used component in Linux called systemd that Ayer believes is "defective by design".
However, others believe disclosing such a bug without first contacting systemd's developers is irresponsible. Ayer was critical of systemd for being overly complex and made the argument that Linux developers have "fallen behind other operating systems in writing secure and robust software".
On 02/11/2017 08:21 AM, Luke Kenneth Casson Leighton wrote:
https://www.theregister.co.uk/2017/01/24/systemd_flaw/
"Newer" versions of systemd deployed by Fedora or Ubuntu have been secured, but Debian systems are still running an older version and therefore need updating.
systemd is a suite for building blocks for Linux systems that provides system and service management technology. Security specialists view it with suspicion and ***>>>complaints about function creep are not uncommon<<<***.
https://betanews.com/2016/10/07/systemd-vulnerability-linux-crash/
The reason he has decided to disclose the bug publicly was to bring further attention to problems with a widely used component in Linux called systemd that Ayer believes is "defective by design".
However, others believe disclosing such a bug without first contacting systemd's developers is irresponsible. Ayer was critical of systemd for being overly complex and made the argument that Linux developers have "fallen behind other operating systems in writing secure and robust software".
This is all FUD. Of course systemd ends up with vulnerabilities because of bugs. So does Linux, Bash, OpenSSL, SSH, Apache, etc. Debian responds to those vulnerabilities by fixing them. There is no fundamental difference with systemd.
If you want to talk about vulnerabilities, a years-old snapshot of Debian Testing is almost certainly *filled* with vulnerabilities all over the place, and only technically minded people will know how to fix them, because this is an old Testing snapshot. So in the name of "ethics" where it's somehow unethical to distribute a 100% libre program you don't like, you'll be giving any non-technical users an insecure system that they don't know how to update, and if they do find out how, they'll just be left wondering why it wasn't updated in the first place. Most likely, they'll assume that you are incompetent or just don't care.
And this is especially bad considering that of all the distros you offered, Debian is the most user-friendly, if you distribute *stable, stock* Debian. That was the only reason why I ordered some Debian cards. Knowing that you are not delivering what I want to be on the card that I'm going to give to my mother, I see now that this was completely pointless. I'm going to have to do all of the work to make sure she has a system she can use properly because you refuse to cooperate just by delivering the current, stable, stock Debian.
This is not something that personally affects me very much; I should be able to figure out how to install Debian on my own, and I was planning to do so anyway. But you are making it needlessly difficult for your project to succeed by taking this zealous hardline stance against systemd; it means that only retailers that know how to install whatever OS the user wants (e.g. Think Penguin) will be able to sell anything that non-technical people can use. You can forget your dream of having EOMA68 hardware on Wal-mart's shelf in that case.
--- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Sat, Feb 11, 2017 at 2:43 PM, Julie Marchant onpon4@riseup.net wrote:
And this is especially bad considering that of all the distros you offered, Debian is the most user-friendly, if you distribute *stable, stock* Debian. That was the only reason why I ordered some Debian cards.
if you misunderstood and believed that you were buying a product, that you were placing an order, as opposed to helping reach the goal of bringing ethically-developed eco-conscious computing devices to mass-volume, then on the basis that i can only accept money from people who are 100% happy with the service that i am providing i am obligated, even though the components have been ordered from the factory, to offer you the opportunity to have your money returned to you. 100% integrity is *that* important to me, it takes *absolute* precedence.
one of
Knowing that you are not delivering what I want to be on the card that I'm going to give to my mother, I see now that this was completely pointless. I'm going to have to do all of the work to make sure she has a system she can use properly because you refuse to cooperate just by delivering the current, stable, stock Debian.
it's actually very simple to do (i wish it was as easy as using debian-installer): either find someone else's rootfs and literally just drop it onto the microsd card, or, if you cannot trust random arbitrary downloads from the internet of 4 gigabytes in size, look up debian qemu or debootstrap foreign architectures (which use qemu in headless mode i believe to run the final pre-preparation steps), and the job's pretty much done in under an hour.
... ah! here you are: https://wiki.debian.org/EmDebian/CrossDebootstrap#QEMU.2Fdebootstrap_approac...
it's really amazingly straightforward.
This is not something that personally affects me very much; I should be able to figure out how to install Debian on my own, and I was planning to do so anyway.
great. it would be very helpful if you could document that process, so that others can benefit and also help you out.
But you are making it needlessly difficult for your project to succeed by taking this zealous hardline stance against systemd;
julie: i'm sleeping for about 12 to 14 hours a day, i've some sort of virus that's affected my health for over 25 years and is increasing in its virulence, i haven't the *time* or energy to be zealous.
it's much simpler than you imagine it to be, and it's down to a pathological systemic flaw in the way that software libre is developed (as a world-wide community). i'm deeply disturbed by the way in which systemd has been developed and deployed: it's unethical in ways that go beyond acceptable boundaries which the actual *software license* simply doesn't cover. i haven't the time or energy to spend on it, and its unethical development and deployment is not something i can endorse or distribute to people who are trusting me to deliver them ethically-developed hardware *and software*.
i don't expect everyone to understand that.
l.
On 02/11/2017 02:53 PM, Luke Kenneth Casson Leighton wrote:
if you misunderstood and believed that you were buying a product, that you were placing an order, as opposed to helping reach the goal of bringing ethically-developed eco-conscious computing devices to mass-volume
Every time someone mentions customers, it seems you bring up this exact same line. I don't care if *I* am a customer or not; you *are* going to have customers, are you not? Or is this just always going to be an endless exchange of "gifts" for you, where you get X orders for this product, abandon it, and move on to the next one?
Unless that's your plan, you're going to need to address people's concerns and complaints at some point; you can't just keep handwaving them away forever. So you're better off addressing them *now*, when the people making the complaints are doing so because they *care*.
100% integrity is *that* important to me, it takes *absolute* precedence.
Also, please, get off your high horse:
1. You're using proprietary CAD software. 2. You use YouTube. 3. You have not released all the PCB CAD files. 4. You have admitted to signing NDAs.
Et cetera.
You are not perfect, you cannot be perfect, and no one expects you to be perfect. Don't make the perfect the enemy of the good.
i haven't the *time* or energy to be zealous.
You're being zealous right now when you say things like this:
i'm deeply disturbed by the way in which systemd has been developed and deployed: it's unethical in ways that go beyond acceptable boundaries which the actual *software license* simply doesn't cover.
And you're so convinced that systemd is the worst thing in the world, for reasons you cannot explain, That you would rather ship a testing snapshot of Debian from years ago than just ship a stable version of stock Debian.
This goes beyond "not having time" to "deal with" systemd. You are *actively* making decisions to "protect" people from what they *specifically* asked for, which you also *specifically* offered.
If you want to talk about ethics, what about the ethics of promising one thing, but delivering another? OK, you're technically delivering Debian, but no one is going to see delivery of a years-old Testing snapshot as delivering on your promise. You'll be seen as a guy who is trying to force anti-systemd hysteria down their throats.
i haven't the time or energy to spend on it
Then why did you offer Debian and Fedora as options? By offering them as options, you implicitly volunteered to *at least* install stock Debian and stock Fedora. You could just as easily have only offered Devuan.
if people - even on this list - did not go into full-on "attack" mode, getting extremely defensive and upset when i try to point something out, instead took the time to ask questions, it would be a lot easier.
It's you who is being defensive, Luke. No one is in "attack mode". You have this bad habit where you treat your critics like they are adversaries. We are not your adversaries. The only reason I am so critical of your decisions is because I want EOMA68 to succeed.
--- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Sun, Feb 12, 2017 at 4:26 AM, Julie Marchant onpon4@riseup.net wrote:
On 02/11/2017 02:53 PM, Luke Kenneth Casson Leighton wrote:
if you misunderstood and believed that you were buying a product, that you were placing an order, as opposed to helping reach the goal of bringing ethically-developed eco-conscious computing devices to mass-volume
Every time someone mentions customers, it seems you bring up this exact same line. I don't care if *I* am a customer or not; you *are* going to have customers, are you not?
no, julie, i'm not. the certification mark process requires that i not have any customers at all, only licensees of the certification mark. the only reason why i'm here at all is to bootstrap the ecosystem, and to fulfil the requirements in order to be able to even *apply* for the certification mark i have to be very *very* careful not to have actual customers.
l.
On 02/12/2017 03:56 AM, Luke Kenneth Casson Leighton wrote:
no, julie, i'm not. the certification mark process requires that i not have any customers at all, only licensees of the certification mark. the only reason why i'm here at all is to bootstrap the ecosystem
Well, that changes everything. I'm sorry I misunderstood; I always assumed that you would start selling the hardware after the first batch was made (like e.g. the OpenPandora).
I still think your opposition to systemd is unfounded and wrong, and I still think that you should drop Debian and Fedora as options if you're going to do it, but as long as it only affects hardcore supporters / early adopters, everything will be fine.
Also, I understand now that the way I have been writing has unintentionally caused distress. That being the case, I apologize.
On Sun, Feb 12, 2017 at 2:37 PM, Julie Marchant onpon4@riseup.net wrote:
On 02/12/2017 03:56 AM, Luke Kenneth Casson Leighton wrote:
no, julie, i'm not. the certification mark process requires that i not have any customers at all, only licensees of the certification mark. the only reason why i'm here at all is to bootstrap the ecosystem
Well, that changes everything. I'm sorry I misunderstood; I always assumed that you would start selling the hardware after the first batch was made (like e.g. the OpenPandora).
i simply can't.... and fulfil the planned aims at the same time. lots of the decisions made make absolutely no sense to those people and companies who have been making (and selling) 5k, 10k, even 50k hardware
Also, I understand now that the way I have been writing has unintentionally caused distress. That being the case, I apologize.
it's ok julie - it happens. goodness knows i've annoyed enough people :)
l.
On Wed, Mar 8, 2017 at 5:54 AM, Luke Kenneth Casson Leighton lkcl@lkcl.net wrote:
it's ok julie - it happens. goodness knows i've annoyed enough people :)
... also, i have to say, i really appreciate (and missed) your insights and feedback. l.
On February 12, 2017 12:56:57 AM PST, Luke Kenneth Casson Leighton lkcl@lkcl.net wrote:
.... the certification mark process requires that i not have any customers at all, only licensees of the certification mark. the only reason why i'm here at all is to bootstrap the ecosystem, and to fulfil the requirements in order to be able to even *apply* for the certification mark i have to be very *very* careful not to have actual customers.
This is an important point. Will it be part of a future update? I was under the impression that rhombus-tech would be producing and selling products *without* the certification mark to jump-start a market where others would produce and apply for the certification mark.
If a different entity out of this community must take over the flagship of the first mass-produced products, I'm confident the research and information developed up to the point of shipping to backers is enough to facilitate that happening. I do however wonder who is going to be doing it.
-- Eric Duhamel http://www.noxbanners.net/
Hallo,
Am 12.02.2017 um 05:26 schrieb Julie Marchant:
Also, please, get off your high horse:
- You're using proprietary CAD software.
- You use YouTube.
- You have not released all the PCB CAD files.
- You have admitted to signing NDAs.
Et cetera.
I will put something on top of it because I am getting angry: Unfortunately Luke is or is becoming a Ethics-Extremist. Extremism allways does harm. So it is hurting the project (and Lukes health). Out of his extremism he uses an extreme language. He calls for example enterprises and people criminal, which never have broken any law. One could write, that have conducted completely wrong, but calling them criminal is burning bridges. This extreme language is one of the reasons for the heated discussion on the list.
If Luke is so extrem about ethics, how can he distribute hardware at all? There is no ethical hardware: they get gold and/or other material from warlords or other criminals, the working conditions are ruining the health of the workers, chip-production does great harm to the enviroment, ...
I cannot see that Luke has tackeld any of this problems at all, which he should as he is so extrem. Distributing hardware is much, much more unethical than distributing systemd.
One last example: Luke hates Fairphone too and calls them a criminal enterprise. Fairphone not Apple! That is burning bridges! Fairphone as a tiny enterprise does try to go a few steps in the direction of ethical hardware. They try to find ethical produced minerals, they care about working conditions. In these questions they are far ahead of Luke (in other questions Luke is far ahead). In his hate Luke cannot respect what they do and does not recognize any progress, because they are great criminals.
So for me there are limits in being ethical. You can be ethical in only part of your life. That should make you refining from extreme positions and not using extrem language. Some are unethical in this area, which you can critizise, you are unethical in an other area. So be kind with the people, which does not exclude sharp critizism, and you would too reduce the harm done to your project (and your health).
Wolfgang
On Sun, Feb 12, 2017 at 3:12 PM, Wolfgang Romey hier@wolfgangromey.de wrote:
I will put something on top of it because I am getting angry: Unfortunately Luke is or is becoming a Ethics-Extremist. Extremism allways does harm.
i've since discovered a more "understandable" version of the bill of ethics, known as the titanian's "code of honour". it boils down to "we do no harm; we always do good; the code applies at all times, no exceptions; everyone knows the code".
sticking to an ethical code 100% is *really tough* wolfgang, even when the code says (in essence), "we do no harm". now, note that it does NOT say "we keep everybody HAPPY". studies have repeatedly shown that using "happiness" as a metric results in pathological group behaviour.
one other aspect of the titanian's code of honour is: constructive feedback should always be sought and, if provided, NEVER denied.
so i appreciate your perspective, wolfgang, but bear in mind that someone who is *aware* of the Bill of Ethics should *never* be "causing harm".
bear in mind, though that people (or entities) who (or which) are not self-aware are *automatically* excluded from the protections of the Bill of Ethics.
So it is hurting the project (and Lukes health). Out of his extremism he uses an extreme language. He calls for example enterprises and people criminal, which never have broken any law. One could write, that have conducted completely wrong, but calling them criminal is burning bridges. This extreme language is one of the reasons for the heated discussion on the list.
apologies, wolfgang, but denial of the facts - even complex ones - doesn't help. i freely admit that i have no idea how to present these things to people without them getting extremely upset. perhaps you could advise on how the perspective that i have should be presented... but bear in mind that neither the analysis nor the conclusions nor the ethics themselves on which i base the perspective that i hold are up for debate or negotiable.
so many people compromise on things that really do matter, it's up to me to say "no" and to *find* a way.
If Luke is so extrem about ethics, how can he distribute hardware at all? There is no ethical hardware: they get gold and/or other material from warlords or other criminals, the working conditions are ruining the health of the workers, chip-production does great harm to the enviroment, ...
*sigh* i know. i can't tackle everything at once, so i had to make a decision on what - at the *current* level of resources - *can* be tackled.
other issues will be tackled later, *once* the power and influence has extended to that level.
One last example: Luke hates Fairphone too
did you know that experiencing hate *literally* causes your body to create toxins with similar chemical properties to snake venom? knowing this to be the case, and not wishing to (literally) poison myself, we may logically conclude your observation to be false.
and calls them a criminal enterprise.
if they've distributed copyright-violating code, and haven't fixed that, then yes. do you deny that they've distributed copyright-violating code in the past?
So for me there are limits in being ethical. You can be ethical in only part of your life.
we all make choices, wolfgang. there are so many people in the world who choose to compromise, who choose an ordinary and "mundane" existence, i am intelligent enough to realise that i would be doing myself and the world a huge dis-service to make the compromises that you are (indirectly) implying.
i'm sure you've heard this hilarious quote before:
the reasonable man adapts himself to the world the unreasonable man adapts the world to himself therefore, all progress depends on the unreasonable man.
after learning that the definition of "reasonable" is literally "reason-able" as in "has the ability to reason", i actually disagree with the _literal_ meaning of this quote, but i still like it anyway and i know what it's getting at.
l.
On 2017-03-08 at 06:16:20 +0000, Luke Kenneth Casson Leighton wrote:
and calls them a criminal enterprise.
if they've distributed copyright-violating code, and haven't fixed that, then yes. do you deny that they've distributed copyright-violating code in the past?
IANAL, but I'm quite sure that most sane legislations cover this kind of copyright violations in civil law, not in criminal law, so no, if they only did that they didn't commit any crime and they can't be called criminal.
They are people who did something wrong (are still doing? I don't care about android phones so I didn't check) and they are liable to pay damages and other consequences, but nobody of the people involved would get a criminal record because of this.
On 2017-02-11 at 13:21:05 +0000, Luke Kenneth Casson Leighton wrote:
https://www.theregister.co.uk/2017/01/24/systemd_flaw/
"Newer" versions of systemd deployed by Fedora or Ubuntu have been secured, but Debian systems are still running an older version and therefore need updating.
Debian backports (when possibile) security fixes to the packages they distribute; a quick check for the CVE listed in that article shows that most debian systems should be fine:
https://security-tracker.debian.org/tracker/CVE-2016-2118
note that security fixes are released through the "<release> (security)" repository and only merged in "<release>" when there is a point update of it (every few months for as long as the release is supported) and most systems do have the security repository enabled (that happens by default with the installer and is considered a good practice).
Wheezy is still listed as vulnerable, but that's because it's out of regular support (since april 2016, currently only under LTS_ support), and thus there won't be another point release to include the changes published via the (security) repo.
.. _LTS: https://wiki.debian.org/LTS
Elena ``of Valhalla'' valhalla-l@trueelena.org writes:
On 2017-02-11 at 13:21:05 +0000, Luke Kenneth Casson Leighton wrote:
https://www.theregister.co.uk/2017/01/24/systemd_flaw/
"Newer" versions of systemd deployed by Fedora or Ubuntu have been secured, but Debian systems are still running an older version and therefore need updating.
Debian backports (when possibile) security fixes to the packages they distribute; a quick check for the CVE listed in that article shows that most debian systems should be fine:
That should be:
https://security-tracker.debian.org/tracker/CVE-2016-10156
Note the "<not-affected>" against all releases.
The "fixed" in the status column actually means "didn't need fixing in the first place" in this instance AFAIK.
Cheers, Phil.
On 2017-02-11 at 17:47:18 +0100, Philip Hands wrote:
That should be:
uops, thanks, big duckduckgoing-fail on my part
(I searched for the CVE number + debian, and didn't check that I was actually opening the right one)
Elena ``of Valhalla'' valhalla-l@trueelena.org writes:
On 2017-02-11 at 17:47:18 +0100, Philip Hands wrote:
That should be:
uops, thanks, big duckduckgoing-fail on my part
(I searched for the CVE number + debian, and didn't check that I was actually opening the right one)
np -- I'd done exactly the same thing earlier and was surprised when then getting to a CVE for Samba -- clearly duckduckgo were trying to be helpful by showing us a vulnerability that actually existed on Debian ;-)
Cheers, Phil.
On Sat, Feb 11, 2017 at 5:40 PM, Elena ``of Valhalla'' valhalla-l@trueelena.org wrote:
On 2017-02-11 at 17:47:18 +0100, Philip Hands wrote:
That should be:
uops, thanks, big duckduckgoing-fail on my part
(I searched for the CVE number + debian, and didn't check that I was actually opening the right one)
thank you to both of you for investigating.
l.
Luke Kenneth Casson Leighton lkcl@lkcl.net writes:
https://www.theregister.co.uk/2017/01/24/systemd_flaw/
"Newer" versions of systemd deployed by Fedora or Ubuntu have been secured, but Debian systems are still running an older version and therefore need updating.
You appear to be in full confirmation-bias mode when it comes to systemd, and thus will accept any criticism of systemd as truth without applying any critical thinking at all.
That makes your signal to noise ratio on this subject _really_ poor.
In this particular case, the vunlerability was in systemd v228.
No release version of Debian has ever shipped that version.
The version in Debian stable is 215-17+deb8u6 -- so was never vulnerable.
That's why there's no DSA (Debian Security Alert) related to this.
Of course, I don't know why I'm bothering to point this out. As I said, confirmation bias means that none of you that despise systemd will take the slightest notice, and I see that recent sociological research shows that doing things like debunking Trump's unusual versions of reality actually hardens the views of his supporters, becuase people are cheerful to assume that the source of the critism is fatally biased, and then spend mental effort on contradicting what is being said by coming up with counter-arguments, which they then remeber for later. *sigh*
Cheers, Phil.
On Sat, Feb 11, 2017 at 4:29 PM, Philip Hands phil@hands.com wrote:
Luke Kenneth Casson Leighton lkcl@lkcl.net writes:
https://www.theregister.co.uk/2017/01/24/systemd_flaw/
"Newer" versions of systemd deployed by Fedora or Ubuntu have been secured, but Debian systems are still running an older version and therefore need updating.
You appear to be in full confirmation-bias mode when it comes to systemd, and thus will accept any criticism of systemd as truth without applying any critical thinking at all.
it's simpler than that, phil: i'm exhausted about 90% of the time, am making mistakes, and haven't the energy to fully research something that i know to be unethical in ways that are really very hard for me to explain.
if people - even on this list - did not go into full-on "attack" mode, getting extremely defensive and upset when i try to point something out, instead took the time to ask questions, it would be a lot easier.
this morning, i visited my host. i hadn't been able to sleep (at all). i woke up on his couch around 11am after he and his family had gone out. i tried to have a conversation: it was as if i was completely drunk. my words were slurred, i couldn't finish sentences, and i couldn't focus my eyes. realising something was wrong i drank two cups of water straight and within *thirty seconds* i was okay. tired, but okay.
much of my life i have been able to *perceive* really serious problems far better than i've been able to *explain* them. i'm amazed to have gotten this far with this campaign and it's down to the fact that so many people have helped out in so many ways. that includes you, most of all, phil.
so if i may ask you something: be kind, please. all of you. i appreciate your honesty, and i appreciate you keeping an eye out to compensate for the mistakes i make. i would however ask all of you to be a little bit kinder in the way that you do that, that's all. don't ever stop being honest and up-front, though. i'd rather you splurged angrily than remain quiet and let a fundamental mistake cascade through by omission.
l.
My 0.02$ Debian, Fedora and Parabola are, to be quite honest, very ethically made distributions. Sure Fedo has *cough*kernelblobs*cough* but that can be fixed very easy by swapping the standard kernel with linux-libre (You should totally do this btw) This is a change I would totally back. Systemd, cancer as it may be developed, is still free software. So as long as I have the freedom to get rid of it, change it, or even keep it, then it isn't worth purging. But please Luke, don't decide for me, I got enough of that from Microsoft with Winblows 8. (You need some tiles in your life son!) One of the absolute biggest appeals of your cards is that they actually can run mainline everything. Please recognize this is a huge huge deal especially on a distros like Debian that actually has official ARM builds. Don't mess with it. Stock xfce Debian please.
On February 11, 2017 11:10:56 PM GMT+03:00, Luke Kenneth Casson Leighton lkcl@lkcl.net wrote:
On Sat, Feb 11, 2017 at 4:29 PM, Philip Hands phil@hands.com wrote:
Luke Kenneth Casson Leighton lkcl@lkcl.net writes:
https://www.theregister.co.uk/2017/01/24/systemd_flaw/
"Newer" versions of systemd deployed by Fedora or Ubuntu have been secured, but Debian systems are still running an older version and therefore need updating.
You appear to be in full confirmation-bias mode when it comes to systemd, and thus will accept any criticism of systemd as truth
without
applying any critical thinking at all.
it's simpler than that, phil: i'm exhausted about 90% of the time, am making mistakes, and haven't the energy to fully research something that i know to be unethical in ways that are really very hard for me to explain.
if people - even on this list - did not go into full-on "attack" mode, getting extremely defensive and upset when i try to point something out, instead took the time to ask questions, it would be a lot easier.
this morning, i visited my host. i hadn't been able to sleep (at all). i woke up on his couch around 11am after he and his family had gone out. i tried to have a conversation: it was as if i was completely drunk. my words were slurred, i couldn't finish sentences, and i couldn't focus my eyes. realising something was wrong i drank two cups of water straight and within *thirty seconds* i was okay. tired, but okay.
much of my life i have been able to *perceive* really serious problems far better than i've been able to *explain* them. i'm amazed to have gotten this far with this campaign and it's down to the fact that so many people have helped out in so many ways. that includes you, most of all, phil.
so if i may ask you something: be kind, please. all of you. i appreciate your honesty, and i appreciate you keeping an eye out to compensate for the mistakes i make. i would however ask all of you to be a little bit kinder in the way that you do that, that's all. don't ever stop being honest and up-front, though. i'd rather you splurged angrily than remain quiet and let a fundamental mistake cascade through by omission.
l.
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
--- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Sat, Feb 11, 2017 at 9:20 PM, Allan Mwenda allanitomwesh@gmail.com wrote:
My 0.02$ Debian, Fedora and Parabola are, to be quite honest, very ethically made distributions.
sorry, allan: that's not the case. if a decision-making process is by "consensus", ignoring or overruling the wishes of *any* one person, then by definition is is unethical. there is a fundamental and in many cases pathologically-held belief that is unfortunately ingrained into western culture that "democracy" - majority voting - is acceptable. the reality is that *by definition* it's fundamentally unethical.
Sure Fedo has *cough*kernelblobs*cough* but that can be fixed very easy by swapping the standard kernel with linux-libre (You should totally do this btw) This is a change I would totally back. Systemd, cancer as it may be developed, is still free software.
the unfortunate thing that i am observing is that just because the license is libre, it doesn't necessarily result in ethical or wise decisions. i've witnessed that continually, now, for almost 20 years.
So as long as I have the freedom to get rid of it, change it, or even keep it, then it isn't worth purging.
do you know the history of hans reiser? is his software available today? given his history, if his software *was* available today, do you think anyone would want to install it and use it? why not? it's free software, right? you can change it, keep it, it's entirely under a libre license... so what's the problem?
But please Luke, don't decide for me,
allan: please do not appeal to me as if i have the right to overrule your own right to self-determination. let me be absolutely clear: i don't appreciate the implication that i have *any* right to make your decisions. i do not. sorry for having to be so blunt.
I got enough of that from Microsoft with Winblows 8. (You need some tiles in your life son!) One of the absolute biggest appeals of your cards is that they actually can run mainline everything.
yeah... except it's incomplete, hence why i will need to still ship with 3.4. people are at liberty to replace that with mainline and see how far they get. i was only able to run up until about 4.7pre1 or so without problems. i compiled over a hundred different kernels around that mark trying to find the problematic patch... ran out of time.
Please recognize this is a huge huge deal especially on a distros like Debian that actually has official ARM builds.
the official arm builds don't have support for EOMA68 because the devicetree overlays aren't part of mainline yet in a version of the mainline linux kernel that's stable on EOMA68-A20 boards.
Don't mess with it.
i don't intend to.
Stock xfce Debian please.
of course it will be stock xfce. of course it will be stock debian - just an older (and thoroughly tested) version.
the line however of systemd is not one that i can cross. you're going to have to demonstrate to me that systemd has been developed (and deployed) in a 100% ethical manner, for that to happen.
now, i have no intention of deciding anything for you. you are entirely at liberty to put whatever OS you choose onto this hardware.
the reason why i made four separate OSes is so that those people who *can't* follow instructions online to set up their own OS have something to get started from.
for those of you who *can* do that i *do* expect you to do so, and, once you've done so, to get together and help out others who can't.
i'll take responsibility for helping those people who need a prebuilt OS. if you have the technical knowledge, i expect you to take responsibility for setting up your own preferred OS.
clear?
l.
sorry, allan: that's not the case. if a decision-making process is by "consensus", ignoring or overruling the wishes of *any* one person, then by definition is is unethical. there is a fundamental and in many cases pathologically-held belief that is unfortunately ingrained into western culture that "democracy" - majority voting - is acceptable. the reality is that *by definition* it's fundamentally unethical.
How can any decision-making process meet this requirement? There are often conflicting desires between different members of a community. In such a case, making *any* decision will have overruled the wishes of at least one community member. Democracy at least lets everyone express their wishes, rather than having e.g. a dictator that makes decisions regardless of the wishes of the people under their jurisdiction.
On Sun, Feb 12, 2017 at 1:37 AM, Jonathan Frederickson silverskullpsu@gmail.com wrote:
How can any decision-making process meet this requirement?
a 20 year study involving hundreds of people was the subject of this very question.
There are often conflicting desires between different members of a community. In such a case, making *any* decision will have overruled the wishes of at least one community member. Democracy at least lets everyone express their wishes,
does that then empower them to have their wishes *met*? do you feel that it's good for people to be able to *express* their wishes if they know that they're not going to be heard?
rather than having e.g. a dictator that makes decisions regardless of the wishes of the people under their jurisdiction.
do you believe that all dictatorships are inherently bad for the people under their jurisdiction? are you familiar with paddy ashdown's instatement as (effective) dictator of one of the (unstable) eastern eurpoean block countries, about ten years ago? he said it was scary as hell, having that much power. the sheer overwhelming responsibility kept him from abusing the power of being head of state, head of the armed forces, judge, jury and executioner, head of the central bank and everything else.
l.
On 02/11/2017 05:51 PM, Luke Kenneth Casson Leighton wrote:
you're going to have to demonstrate to me that systemd has been developed (and deployed) in a 100% ethical manner
So systemd is guilty until proven innocent?
Alright then, unless you can demonstrate to me that the EOMA68-A20 has been developed and deployed in a 100% "ethical" manner (where I am not going to explain what "ethical" means), I am going to assume that it is unethical and refuse to support it.
Except I'm not actually going to do that, because that would be unreasonable. There's a reason our courts don't work this way.
Julie, while I appreciate your efforts at convincing our kind leader here to change his mind, you're not going to make any headway. And, for the record, one of the reasons that Luke has trouble explaining things is because he has Asperger's Syndrome, a form of Autism and a disorder primarily affecting communication and social skills. I have Asperger's as well, and so I can appreciate how it sometimes renders one speechless, or forces one to use the wrong words for things -- often at a sensitive time.
It's unbelievably frustrating.
In the meantime, since the metaphorical horse is very dead at this point, perhaps you should stop beating it. I'm sorry if I sound crass -- but you've voiced your opinion quite thoroughly at this point. I don't see much purpose in continuing to do so -- for sure there's no confusion as to your position. Your opinion matters, and it was heard -- that's all you need to do. That's all you really /can/ do -- and that should be obvious by now. You're not going to win anything (even an argument) by persisting, and it's not really your job to do so anyways. Luke is the one who makes the decisions. He's done that -- he's made his decision. That's basically all there is to it. All you are doing now is making noise, and it hurts my head and I'd like it to stop before I have to reach for the bottle of ibuprofen.
Luke's made his bed, for better or worse. Let him lie in it. Quietly. You are thanked in advance.
On Sun, Feb 12, 2017 at 4:53 AM, Christopher Havel laserhawk64@gmail.com wrote:
Julie, while I appreciate your efforts at convincing our kind leader here to change his mind, you're not going to make any headway. And, for the record, one of the reasons that Luke has trouble explaining things is because he has Asperger's Syndrome, a form of Autism and a disorder primarily affecting communication and social skills. I have Asperger's as well, and so I can appreciate how it sometimes renders one speechless, or forces one to use the wrong words for things -- often at a sensitive time.
It's unbelievably frustrating.
... would you believe it, my conversations with dr stallman cause *me* to have to take deep breaths on a regular basis.
chris, i appreciate your insight (from personal experience) here. your phrasing is, in places, as bad as mine can be, but i know your heart's in the right place.
it is a recurring theme from interactions with people around me that they in effect ask "give me ONE good reason why i should listen to you", and i simply... can't. the reason is: my brain simply doesn't work that way.
in researching why that is, i came across something called "demster shafer theory". it's a generalisation of bayes theorem, and i was interested in it as a way to work out *why* i was good at reverse-engineering (from my work on samba) but also to find out if there was a way to *improve* my ability as a reverse-engineer and knowledge derivation expert.
demster-shafer theory basically says that you may statistically derive a result by taking two *independent* variables in a massively-complex field, work out the probability of them occurring together (independently) then you are permitted to *REPEAT* that exercise and to *SUM* the resultant totally independent results as a way to gain a statistically-valid result across the ENTIRE FIELD.
in this context, the question "give me ONE good reason" is a completely INVALID one.
hence, can you (all) understand that if you ask me "give me ONE good reason" i LITERALLY cannot do that. i could however give you about a hundred SMALLER reasons each with a low statistical probability of them occurring.
... but it would overwhelm you to do so, you would (as you have clearly done so) REJECT the entire APPROACH that i've taken because it's NOT SOMETHING YOU UNDERSTAND AS BEING VALID.
i come across this time and time again, in the physics forums i'm on, on free software mailing lists, at workplaces where i can tell there's something deeply wrong from a whole stack of clues but i CAN'T VOCALISE THEM ALL.
my brain *literally* works in a completely different way from most people's on the planet, in a massively-parallel statistical-inference fashion that hugely and rapidly short-cuts areas of avenue that would cause most people to get stuck and waste months to decades of their life investigating to no avail... but this type of approach is NOT what the human brain was designed to do, and it comes with a heavy penalty both for my health but also in terms of making it REALLY difficult to justify the conclusions (or "intuitions") that my brain flags up as being so brightly "red" that i can no more ignore them than i could if they were say actual threats on my life.
i would *really* appreciate your patience on this. knowing what i do about myself, i deliberately tackle areas that nobody else does. unfortunately, what's happened in the past is that people have stolen the results and the credit for the work that i've done. did you know for example that the openchange project's success is down to my work (not theirs) in reverse-engineering exchange 5.5 back around 2003? of course you don't... because after tracking my research continually they DIDN'T MENTION WHERE THEY GOT THE INFORMATION FROM.
that means that you BELIEVE i am not worth respecting, because my name is not up there in neon lights next to those of "linus torvalds" or "eric raymond" or "bruce perens" or any other person you've heard of and respect in software libre for their achievements.
i act in the background, tackling the things that these people *CAN'T* understand and, because of their position, couldn't deal with anyway because they now have too much responsibility in their chosen field of expertise and endeavour to consider abandoning the people who now depend on them.
put another way:
systemd has a huge - MASSIVE - series of independent statistical correlations associated with it, none of them INDIVIDUALLY being statistically significant or indicative of anything (because they're independent events) but when added up overall, using demster-shafer theory, give support for the hypothesis that there is something deeply, deeply wrong with systemd with a confidence level somewhere around 4 sigma. i simply cannot ignore that, but equally i cannot really explain it in ways that you would ACCEPT, either, because my name is not "linus torvalds" or "dr richard stallman".
anyway. the fact that the 3.4 kernel has to be used makes it entirely moot. which reminds me the last time this happened, was when i was working in portsmouth, something similar happened. i was ORDERED to deploy ubuntu, but could not explain or vocalise the dozens of reasons why that was a bad idea. finally one of the sysadmins got fed up of hearing the discussion, did some research and found that canonical had long since terminated support for 486 processors.
so please. understand. sometimes i *can't give you a concrete reason* because there are instead potentially *hundreds* of lower-probability ones, some of which i'm not even consciously aware of.
l.
I don't think effort should be put into purging systemd.
On February 12, 2017 2:10:07 PM GMT+03:00, Luke Kenneth Casson Leighton lkcl@lkcl.net wrote:
On Sun, Feb 12, 2017 at 4:53 AM, Christopher Havel laserhawk64@gmail.com wrote:
Julie, while I appreciate your efforts at convincing our kind leader
here to
change his mind, you're not going to make any headway. And, for the
record,
one of the reasons that Luke has trouble explaining things is because
he has
Asperger's Syndrome, a form of Autism and a disorder primarily
affecting
communication and social skills. I have Asperger's as well, and so I
can
appreciate how it sometimes renders one speechless, or forces one to
use the
wrong words for things -- often at a sensitive time.
It's unbelievably frustrating.
... would you believe it, my conversations with dr stallman cause *me* to have to take deep breaths on a regular basis.
chris, i appreciate your insight (from personal experience) here. your phrasing is, in places, as bad as mine can be, but i know your heart's in the right place.
it is a recurring theme from interactions with people around me that they in effect ask "give me ONE good reason why i should listen to you", and i simply... can't. the reason is: my brain simply doesn't work that way.
in researching why that is, i came across something called "demster shafer theory". it's a generalisation of bayes theorem, and i was interested in it as a way to work out *why* i was good at reverse-engineering (from my work on samba) but also to find out if there was a way to *improve* my ability as a reverse-engineer and knowledge derivation expert.
demster-shafer theory basically says that you may statistically derive a result by taking two *independent* variables in a massively-complex field, work out the probability of them occurring together (independently) then you are permitted to *REPEAT* that exercise and to *SUM* the resultant totally independent results as a way to gain a statistically-valid result across the ENTIRE FIELD.
in this context, the question "give me ONE good reason" is a completely INVALID one.
hence, can you (all) understand that if you ask me "give me ONE good reason" i LITERALLY cannot do that. i could however give you about a hundred SMALLER reasons each with a low statistical probability of them occurring.
... but it would overwhelm you to do so, you would (as you have clearly done so) REJECT the entire APPROACH that i've taken because it's NOT SOMETHING YOU UNDERSTAND AS BEING VALID.
i come across this time and time again, in the physics forums i'm on, on free software mailing lists, at workplaces where i can tell there's something deeply wrong from a whole stack of clues but i CAN'T VOCALISE THEM ALL.
my brain *literally* works in a completely different way from most people's on the planet, in a massively-parallel statistical-inference fashion that hugely and rapidly short-cuts areas of avenue that would cause most people to get stuck and waste months to decades of their life investigating to no avail... but this type of approach is NOT what the human brain was designed to do, and it comes with a heavy penalty both for my health but also in terms of making it REALLY difficult to justify the conclusions (or "intuitions") that my brain flags up as being so brightly "red" that i can no more ignore them than i could if they were say actual threats on my life.
i would *really* appreciate your patience on this. knowing what i do about myself, i deliberately tackle areas that nobody else does. unfortunately, what's happened in the past is that people have stolen the results and the credit for the work that i've done. did you know for example that the openchange project's success is down to my work (not theirs) in reverse-engineering exchange 5.5 back around 2003? of course you don't... because after tracking my research continually they DIDN'T MENTION WHERE THEY GOT THE INFORMATION FROM.
that means that you BELIEVE i am not worth respecting, because my name is not up there in neon lights next to those of "linus torvalds" or "eric raymond" or "bruce perens" or any other person you've heard of and respect in software libre for their achievements.
i act in the background, tackling the things that these people *CAN'T* understand and, because of their position, couldn't deal with anyway because they now have too much responsibility in their chosen field of expertise and endeavour to consider abandoning the people who now depend on them.
put another way:
systemd has a huge - MASSIVE - series of independent statistical correlations associated with it, none of them INDIVIDUALLY being statistically significant or indicative of anything (because they're independent events) but when added up overall, using demster-shafer theory, give support for the hypothesis that there is something deeply, deeply wrong with systemd with a confidence level somewhere around 4 sigma. i simply cannot ignore that, but equally i cannot really explain it in ways that you would ACCEPT, either, because my name is not "linus torvalds" or "dr richard stallman".
anyway. the fact that the 3.4 kernel has to be used makes it entirely moot. which reminds me the last time this happened, was when i was working in portsmouth, something similar happened. i was ORDERED to deploy ubuntu, but could not explain or vocalise the dozens of reasons why that was a bad idea. finally one of the sysadmins got fed up of hearing the discussion, did some research and found that canonical had long since terminated support for 486 processors.
so please. understand. sometimes i *can't give you a concrete reason* because there are instead potentially *hundreds* of lower-probability ones, some of which i'm not even consciously aware of.
l.
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
On February 12, 2017 3:10:07 AM PST, Luke Kenneth Casson Leighton lkcl@lkcl.net wrote:
On Sun, Feb 12, 2017 at 4:53 AM, Christopher Havel laserhawk64@gmail.com wrote:
Julie, while I appreciate your efforts at convincing our kind leader
here to
change his mind, you're not going to make any headway. And, for the
record,
one of the reasons that Luke has trouble explaining things is because
he has
Asperger's Syndrome, a form of Autism and a disorder primarily
affecting
communication and social skills. I have Asperger's as well, and so I
can
appreciate how it sometimes renders one speechless, or forces one to
use the
wrong words for things -- often at a sensitive time.
It's unbelievably frustrating.
... would you believe it, my conversations with dr stallman cause *me* to have to take deep breaths on a regular basis.
chris, i appreciate your insight (from personal experience) here. your phrasing is, in places, as bad as mine can be, but i know your heart's in the right place.
it is a recurring theme from interactions with people around me that they in effect ask "give me ONE good reason why i should listen to you", and i simply... can't. the reason is: my brain simply doesn't work that way.
in researching why that is, i came across something called "demster shafer theory". it's a generalisation of bayes theorem, and i was interested in it as a way to work out *why* i was good at reverse-engineering (from my work on samba) but also to find out if there was a way to *improve* my ability as a reverse-engineer and knowledge derivation expert.
demster-shafer theory basically says that you may statistically derive a result by taking two *independent* variables in a massively-complex field, work out the probability of them occurring together (independently) then you are permitted to *REPEAT* that exercise and to *SUM* the resultant totally independent results as a way to gain a statistically-valid result across the ENTIRE FIELD.
in this context, the question "give me ONE good reason" is a completely INVALID one.
hence, can you (all) understand that if you ask me "give me ONE good reason" i LITERALLY cannot do that. i could however give you about a hundred SMALLER reasons each with a low statistical probability of them occurring.
... but it would overwhelm you to do so, you would (as you have clearly done so) REJECT the entire APPROACH that i've taken because it's NOT SOMETHING YOU UNDERSTAND AS BEING VALID.
i come across this time and time again, in the physics forums i'm on, on free software mailing lists, at workplaces where i can tell there's something deeply wrong from a whole stack of clues but i CAN'T VOCALISE THEM ALL.
my brain *literally* works in a completely different way from most people's on the planet, in a massively-parallel statistical-inference fashion that hugely and rapidly short-cuts areas of avenue that would cause most people to get stuck and waste months to decades of their life investigating to no avail... but this type of approach is NOT what the human brain was designed to do, and it comes with a heavy penalty both for my health but also in terms of making it REALLY difficult to justify the conclusions (or "intuitions") that my brain flags up as being so brightly "red" that i can no more ignore them than i could if they were say actual threats on my life.
i would *really* appreciate your patience on this. knowing what i do about myself, i deliberately tackle areas that nobody else does. unfortunately, what's happened in the past is that people have stolen the results and the credit for the work that i've done. did you know for example that the openchange project's success is down to my work (not theirs) in reverse-engineering exchange 5.5 back around 2003? of course you don't... because after tracking my research continually they DIDN'T MENTION WHERE THEY GOT THE INFORMATION FROM.
that means that you BELIEVE i am not worth respecting, because my name is not up there in neon lights next to those of "linus torvalds" or "eric raymond" or "bruce perens" or any other person you've heard of and respect in software libre for their achievements.
i act in the background, tackling the things that these people *CAN'T* understand and, because of their position, couldn't deal with anyway because they now have too much responsibility in their chosen field of expertise and endeavour to consider abandoning the people who now depend on them.
put another way:
systemd has a huge - MASSIVE - series of independent statistical correlations associated with it, none of them INDIVIDUALLY being statistically significant or indicative of anything (because they're independent events) but when added up overall, using demster-shafer theory, give support for the hypothesis that there is something deeply, deeply wrong with systemd with a confidence level somewhere around 4 sigma. i simply cannot ignore that, but equally i cannot really explain it in ways that you would ACCEPT, either, because my name is not "linus torvalds" or "dr richard stallman".
anyway. the fact that the 3.4 kernel has to be used makes it entirely moot. which reminds me the last time this happened, was when i was working in portsmouth, something similar happened. i was ORDERED to deploy ubuntu, but could not explain or vocalise the dozens of reasons why that was a bad idea. finally one of the sysadmins got fed up of hearing the discussion, did some research and found that canonical had long since terminated support for 486 processors.
so please. understand. sometimes i *can't give you a concrete reason* because there are instead potentially *hundreds* of lower-probability ones, some of which i'm not even consciously aware of.
I think it's possible for people to learn to understand and trust your intuition. After all, a lot of good leaders make decisions based on uncanny intuition.
-- Eric Duhamel http://www.noxbanners.net/
--- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Sun, Feb 12, 2017 at 4:36 AM, Julie Marchant onpon4@riseup.net wrote:
On 02/11/2017 05:51 PM, Luke Kenneth Casson Leighton wrote:
you're going to have to demonstrate to me that systemd has been developed (and deployed) in a 100% ethical manner
So systemd is guilty until proven innocent?
it's been demonstrated already to have been unethically developed and deployed (the problem being in this conversation that any time i mention why i believe that to be the case, it's dismissed or rejected - often violently. these reactions *being* symptomatic of the very underlying reason and cause of my deeply-felt concern... concern that i am having a lot of difficulty expressing)
i was inviting people to evaluate that for themselves, for two reasons: firstly i'm completely exhausted so don't have time or energy to go over it, and secondly, if i provide all the answers that removes all opportunity and possibility for learning.
an ethical act is defined as "increasing the truth, love, awareness or creativity (those qualities being synonymous guide-words for the same underlying concept) of one or more people without reducing the same qualities for *anyone*".
now, before asking the crucial questions, i have to ask people to consider answering them *without judgement*, and *without implied or expressed criticism*, in an *objective* fashion. if i ask questions and people react, "what are you saying???? are you saying that debian's process is incompetent??? how DARE you!!!" or much worse reactions than that, we're not going to get anywhere.
we know that there's an underlying systemic breakdown that is *not anyone's fault*. reacting badly by assuming that any *discussion* of the underlying systemic breakdown *is* one particular group or individual's fault is itself part of the problem.
so please don't do it. not on this list.
so. the question is: what behaviours or decisions in the development and deployment of systemd can be shown to have caused a reduction of truth, love, awareness or creativity in at least one person?
put another way (in terms of energy expenditure and resources, which is a parallel and identical way to ask the exact same question):
has any individual, anywhere in the world, had to spend extensive amounts of time and energy as a result of (1) the development or (2) the deployment of systemd, which *could* have been avoided and, because it was not, resulted in a reduction of or diversion of their resources, time or energy?
Alright then, unless you can demonstrate to me that the EOMA68-A20 has been developed and deployed in a 100% "ethical" manner (where I am not going to explain what "ethical" means),
julie, sorry, but that's not reasonable or rational. reading the lines i believe i understand the point you are trying to make. bear in mind i didn't come across an actual definition of an "ethical act" until about 6 months ago, which doesn't help, much of what i've been *trying* to do has been subconscious or just not possible to properly express.
so,
I am going to assume that it is unethical and refuse to support it.
Except I'm not actually going to do that, because that would be unreasonable.
un-reason-able. you can think of a good *reason* why such action would cause a reduction of truth/love/awareness/creativity. you operate by an ethical code. this is fantastic.
There's a reason our courts don't work this way.
interestingly dutch law has a fundamental basis that if someone may show something to be "unreasonable", it is acceptable and *overrides* existing laws. first country i ever heard of that has such a basis.
l.
Can we all just calm down, about the systemd thing.
Luke is trying his hardest, is sick and also KEEP IN MIND, HE DID NOT HAVE TO START THIS CAMPAIGN! WE NEED TO BE MORE GRATEFUL!
I must admit, I don't understand the systemd arguments, but if he is set on doing it that way, then its not a big issue,
we can easily do what we want if need be anyways. IF your on this mailing list, its a good chance you can reinstall if need be.
I myself would not care if he didn't have systemd in by default. All I know is, he is a good ethical person who needs support not endless criticisms.
Please, let us be more patient and kindhearted. He is doing us a favor even doing this campaign.
I still am curious though, and this question is for Luke,
is systemd lack security, privacy or stability?
I just would like Him to answer and no one else to understand his thinking process.
Anyways though not here to condemn anyone, just try to be reasonable that is all.
On 02/12/2017 04:19 AM, Luke Kenneth Casson Leighton wrote:
crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Sun, Feb 12, 2017 at 4:36 AM, Julie Marchant onpon4@riseup.net wrote:
On 02/11/2017 05:51 PM, Luke Kenneth Casson Leighton wrote:
you're going to have to demonstrate to me that systemd has been developed (and deployed) in a 100% ethical manner
So systemd is guilty until proven innocent?
it's been demonstrated already to have been unethically developed and deployed (the problem being in this conversation that any time i mention why i believe that to be the case, it's dismissed or rejected
- often violently. these reactions *being* symptomatic of the very
underlying reason and cause of my deeply-felt concern... concern that i am having a lot of difficulty expressing)
i was inviting people to evaluate that for themselves, for two reasons: firstly i'm completely exhausted so don't have time or energy to go over it, and secondly, if i provide all the answers that removes all opportunity and possibility for learning.
an ethical act is defined as "increasing the truth, love, awareness or creativity (those qualities being synonymous guide-words for the same underlying concept) of one or more people without reducing the same qualities for *anyone*".
now, before asking the crucial questions, i have to ask people to consider answering them *without judgement*, and *without implied or expressed criticism*, in an *objective* fashion. if i ask questions and people react, "what are you saying???? are you saying that debian's process is incompetent??? how DARE you!!!" or much worse reactions than that, we're not going to get anywhere.
we know that there's an underlying systemic breakdown that is *not anyone's fault*. reacting badly by assuming that any *discussion* of the underlying systemic breakdown *is* one particular group or individual's fault is itself part of the problem.
so please don't do it. not on this list.
so. the question is: what behaviours or decisions in the development and deployment of systemd can be shown to have caused a reduction of truth, love, awareness or creativity in at least one person?
put another way (in terms of energy expenditure and resources, which is a parallel and identical way to ask the exact same question):
has any individual, anywhere in the world, had to spend extensive amounts of time and energy as a result of (1) the development or (2) the deployment of systemd, which *could* have been avoided and, because it was not, resulted in a reduction of or diversion of their resources, time or energy?
Alright then, unless you can demonstrate to me that the EOMA68-A20 has been developed and deployed in a 100% "ethical" manner (where I am not going to explain what "ethical" means),
julie, sorry, but that's not reasonable or rational. reading the lines i believe i understand the point you are trying to make. bear in mind i didn't come across an actual definition of an "ethical act" until about 6 months ago, which doesn't help, much of what i've been *trying* to do has been subconscious or just not possible to properly express.
so,
I am going to assume that it is unethical and refuse to support it.
Except I'm not actually going to do that, because that would be unreasonable.
un-reason-able. you can think of a good *reason* why such action would cause a reduction of truth/love/awareness/creativity. you operate by an ethical code. this is fantastic.
There's a reason our courts don't work this way.
interestingly dutch law has a fundamental basis that if someone may show something to be "unreasonable", it is acceptable and *overrides* existing laws. first country i ever heard of that has such a basis.
l.
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
On 12/02/17 16:26, zap wrote:
Can we all just calm down, about the systemd thing.
Agree - it's easy to criticise and hard to make. Luke is doing the jobs of a whole team of designers. People will be able to install what they want on the hardware. I suggest that getting it together is by far the most important thing at this stage.
Best wishes to all,
I second (third? thank you Boris) Zap's call as well. Julie -- I appreciate your apology, regardless of for whom it is meant. For the future: there is a difference between what you can call a "wise endurance" and a "foolish persistence" -- but the line between them is far thinner than I think most folks would like to think. Something to meditate on :)
Wolfgang, perhaps you should not post while angry -- it sounds a bit like you yourself are trying to start something, and that's not cool.
As for Luke. I am blown away by his description of how his mind works -- and, yet, it reminds me somewhat of the gifts of a lot of people I know and know about. It's kind of ironic how, seemingly, everyone's gift is both a blessing and a curse (so to speak). In particular, he reminds me of Cassandra of Troy, a close relative of the (slightly) more famous Helen. Ancient Greek mythos has it that Apollo wanted Cassandra to be his, er, lady for a night, and she agreed and then had second thoughts... so he gave her the ability to see the future -- except that she could never convince anyone of it! I sense that Luke's mind gives him a similar struggle -- he can tell you that something is bad (or not), but heaven help the both of you if you must know why...
It's both fascinating and tragic at the same time.
On Sun, Feb 12, 2017 at 10:36 AM, Boris Barbour boris.barbour@ens.fr wrote:
On 12/02/17 16:26, zap wrote:
Can we all just calm down, about the systemd thing.
Agree - it's easy to criticise and hard to make. Luke is doing the jobs of a whole team of designers. People will be able to install what they want on the hardware. I suggest that getting it together is by far the most important thing at this stage.
Best wishes to all,
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
Oh, whoops, I top-posted again. Apologies to Luke. I know that's one of your pet peeves.
On Sun, Feb 12, 2017 at 3:26 PM, zap zapper@openmailbox.org wrote:
I still am curious though, and this question is for Luke,
is systemd lack security, privacy or stability?
that and more. just one of the big concerns is clear violation of ISO9001 QA "software assurance" standards (also known as "scope creep"). it's... too complex and too comprehensive to go into in one sentence / message, zap.
I just would like Him to answer and no one else to understand his thinking process.
i'm planning to do a special update about it, not least precisely because people are missing (or misreading, or forgetting) things, particularly as newer people come in and don't have the full context.
that then makes it easier for me to refer people and get them up-to-speed, also encourages feedback.
l.
On 2017-02-11 at 22:51:57 +0000, Luke Kenneth Casson Leighton wrote:
do you know the history of hans reiser? is his software available today? given his history, if his software *was* available today, do you think anyone would want to install it and use it? why not? it's free software, right? you can change it, keep it, it's entirely under a libre license... so what's the problem?
Actually, yes, his software *is* available today: according to wikipedia (and to a quick ``ls /lib/modules/4.9.0-1-amd64/kernel/fs/reiserfs/``) ReiserFS is still available in the linux kernel, altought probably used more to access legacy systems than for new deployments. Of course, by this time it's no longer "his software", but more software that belongs to the community, based on code written by Hans Reiser but then modified by multiple people in the meanwhile.
I strongly suspect that the fact that it is an early 2000s filesystem with no modern features is more relevant to the fact that very few people are using it today than the private life of its main (but not only) developer. IIRC its decline had already started before the murder / arrest, because developement on it had already stopped, and new features where being added to Reiser4 (which was never ready for acceptance in mainline, however).
Luke Kenneth Casson Leighton:
and i simply don't have the time - or importantly the energy - to create a new image, *especially* based on people's comments and reactions that they'd be deeply unhappy with it not being a "stock image", even if all i did was make it boot sysvinit instead by default. those comments *alone* immediately terminate all and any possibility that i can provide debian/jessie in a 100% ethical way.
on receipt of the cards, anybody who wants to will be free and entirely at liberty to do "apt-get dist-upgrade" and they will be taking direct responsibility for doing so. for those people who are technically-minded they are also entirely free and at liberty to set up a from-scratch root filesystem.
Great, and what about newbies who want stock image and don't know the nooks and crannies of GNU/Linux?
On February 11, 2017 7:10:00 AM PST, Lyberta lyberta@lyberta.net wrote:
Great, and what about newbies who want stock image and don't know the nooks and crannies of GNU/Linux?
Depending on what you mean by "newbies", I don't think they would know if they want any particular image of GNU/Linux except the one that was designed to run on the product by the maker of the product. After all, they don't know the nooks and crannies of GNU/Linux and would just want to receive a product that works.
-- Eric Duhamel http://www.noxbanners.net/
Eric Duhamel:
Depending on what you mean by "newbies", I don't think they would know if they want any particular image of GNU/Linux except the one that was designed to run on the product by the maker of the product. After all, they don't know the nooks and crannies of GNU/Linux and would just want to receive a product that works.
OK, I'm a software developer, not a system administrator. I have no idea what 90% of packages installed on my system do. I only use terminal to upload my code to the Git repository. But I need g++ 6.
Debian Jessie has g++ 4.9.2 which is extremely old and none of my software will compile there. When I pledged for Debian card, I expected stock Debian with maybe a few custom packages which would be explicitly marked as such. And the first thing I'd do is to upgrade to Testing which as of writing this has g++ 6.3.0.
Now I'm told that issuing "apt-get dist-upgrade" is taking responsibility, etc. So I'm stuck with old and unusable frankendistro and on my own if I want to make it work.
I've chosen GNU/Linux because of its freedom and I've chosen Debian because it doesn't have proprietary software in main. I have no idea of what is going under the hood. I don't care what init system I run as long as it is free software and it boots my PC.
A couple of years ago I needed to buy a laptop. I've looked for one which doesn't come with Windows and I've found one with Ubuntu. Now, I really hate Ubuntu and the first thing I've done was to install Debian in dual boot. For some reason, Debian couldn't power off my laptop so I removed it and I'm still stuck with Ubuntu.
I know what you're going to say: "You should've looked for the solution on the Internet.". Well, sometimes I don't have time and am scared of bricking my hardware. I want things to "just work".
It looks like EOMA68-A20 is not going to just work. I don't want it to end like my laptop.
On February 12, 2017 11:35:00 AM PST, Lyberta lyberta@lyberta.net wrote:
It looks like EOMA68-A20 is not going to just work. I don't want it to end like my laptop.
This is a legit problem. While the first EOMA68-A20 cards need to be shipped without delay, we also NEED to get EOMA computer cards to work with stock Debian and more as soon as possible.
Thankfully it seems the main argument started by this thread has wound down, but I don't know how to properly split off threads into a new subject. If there's anyone who can continue the discussion on getting modern, update-able images on the A20 or other, please do so.
-- Eric Duhamel http://www.noxbanners.net/
Hi,
Le Sun, 12 Feb 2017 11:58:52 -0800 Eric Duhamel ericxdu23@gmail.com a écrit:
[...] but I don't know how to properly split off threads into a new subject [...]
Just start a reply and change the subject in your reply, or if you can't edit the subject in a reply, just post a new message, not a reply, with the intended new subject.
Amicalement,
Starting this new topic on Eric's request. I do find this a priority as shipping an aged Debian/Fedora/etc is not ideal at all. Let's discuss, and maybe even set up targets in the wiki?
On February 12, 2017 10:58:52 PM GMT+03:00, Eric Duhamel ericxdu23@gmail.com wrote:
On February 12, 2017 11:35:00 AM PST, Lyberta lyberta@lyberta.net wrote:
It looks like EOMA68-A20 is not going to just work. I don't want it to end like my laptop.
This is a legit problem. While the first EOMA68-A20 cards need to be shipped without delay, we also NEED to get EOMA computer cards to work with stock Debian and more as soon as possible.
Thankfully it seems the main argument started by this thread has wound down, but I don't know how to properly split off threads into a new subject. If there's anyone who can continue the discussion on getting modern, update-able images on the A20 or other, please do so.
-- Eric Duhamel http://www.noxbanners.net/
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
Hello,
On Mon, Feb 13, 2017 at 7:18 AM, Allan Mwenda allanitomwesh@gmail.com wrote:
Starting this new topic on Eric's request. I do find this a priority as shipping an aged Debian/Fedora/etc is not ideal at all. Let's discuss, and maybe even set up targets in the wiki?
Kernel looks almost OK: http://linux-sunxi.org/Linux_mainlining_effort#Status_Matrix
even if there are still some WIP entries and some green but "?", does anyone tested those ?
I think DRM display (HDMI, etc.) & VE (video Engine) are important, as a lot of people will want them.
There's a section here: http://linux-sunxi.org/index.php?title=Buying_guide%C2%A7ion=6#Rhombus_Tech
Maybe a status page could be added here: http://linux-sunxi.org/Category:Devices
On Mon, Feb 13, 2017 at 09:18:18AM +0300, Allan Mwenda wrote:
Starting this new topic on Eric's request. I do find this a priority as shipping an aged Debian/Fedora/etc is not ideal at all. Let's discuss, and maybe even set up targets in the wiki?
Obligatory link: http://linux-sunxi.org/Linux_mainlining_effort
On February 13, 2017 1:50:29 AM PST, Tzafrir Cohen tzafrir@cohens.org.il wrote:
Obligatory link: http://linux-sunxi.org/Linux_mainlining_effort
Thanks. I wonder if this topic been raised on the mailing list already; I suppose it's a little premature since the initial run of dev cards haven't gotten to backers yet, and we weed hackers to have them in order to start hacking on the kernel.
-- Eric Duhamel http://www.noxbanners.net/
--- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Tue, Feb 14, 2017 at 2:40 AM, Eric Duhamel ericxdu23@gmail.com wrote:
On February 13, 2017 1:50:29 AM PST, Tzafrir Cohen tzafrir@cohens.org.il wrote:
Obligatory link: http://linux-sunxi.org/Linux_mainlining_effort
Thanks. I wonder if this topic been raised on the mailing list already; I suppose it's a little premature since the initial run of dev cards haven't gotten to backers yet, and we weed hackers to have them in order to start hacking on the kernel.
i'm just talking to my host, here in taiwan, to use his equipment to get 50 boards made up. i'll also get some RS232-UARTs and some micro-desktops made as well... or perhaps some breakout boards and the OTG-Host cables, have to see.
any takers please put alias down here so i get a quick idea of numbers http://rhombus-tech.net/allwinner_a10/
l.
On Mon, Feb 13, 2017 at 6:18 AM, Allan Mwenda allanitomwesh@gmail.com wrote:
Starting this new topic on Eric's request. I do find this a priority as shipping an aged Debian/Fedora/etc is not ideal at all. Let's discuss, and maybe even set up targets in the wiki?
please do, pick somewhere (under or at http://rhombus-tech.net/allwinner_a10/) and i'll move it later if needed.
key:
* single monolithic devicetree must NOT repeat NOT go into mainline * it is IMPERATIVE that people DO NOT get the impression that this is "yet another fuckwit SBC monolithic design to be thrown out in only a few months". * dtb files MUST be ONLY for the Card(s) themselves and NOT repeat NOT include the housing peripherals * housing peripherals MUST be in SEPARATE devicetree overlays, one per housing * BOTH uboot AND linux kernel MUST read the I2C EEPROM at address 0x51 to determine which overlay is to be loaded.
* 4.7rc0-4.7rc5 bug has to be found. it's still present in all mainline kernels since (appx) 4.7rc4 and terminates all and any possibility of deployment of mainline linux.
of course, anyone wishing to take on this work will need Cards and Housings to do so. which is why the Cards are going out with linux 3.4 and why i've deliberately kept the numbers down for this first run.
l.
On Sun, 12 Feb 2017 19:35:00 +0000, Lyberta wrote:
Eric Duhamel:
Depending on what you mean by "newbies", I don't think they would know if they want any particular image of GNU/Linux except the one that was designed to run on the product by the maker of the product. After all, they don't know the nooks and crannies of GNU/Linux and would just want to receive a product that works.
OK, I'm a software developer, not a system administrator. I have no idea what 90% of packages installed on my system do. I only use terminal to upload my code to the Git repository. But I need g++ 6.
Debian Jessie has g++ 4.9.2 which is extremely old and none of my software will compile there. When I pledged for Debian card, I expected stock Debian with maybe a few custom packages which would be explicitly marked as such. And the first thing I'd do is to upgrade to Testing which as of writing this has g++ 6.3.0.
Now I'm told that issuing "apt-get dist-upgrade" is taking responsibility, etc. So I'm stuck with old and unusable frankendistro and on my own if I want to make it work.
I've chosen GNU/Linux because of its freedom and I've chosen Debian because it doesn't have proprietary software in main. I have no idea of what is going under the hood. I don't care what init system I run as long as it is free software and it boots my PC.
A couple of years ago I needed to buy a laptop. I've looked for one which doesn't come with Windows and I've found one with Ubuntu. Now, I really hate Ubuntu and the first thing I've done was to install Debian in dual boot. For some reason, Debian couldn't power off my laptop so I removed it and I'm still stuck with Ubuntu.
I know what you're going to say: "You should've looked for the solution on the Internet.". Well, sometimes I don't have time and am scared of bricking my hardware. I want things to "just work".
If you can boot your machine from USB, you won't brick it by changing the OS, because you can boot a Debian or Ubuntu installer and reinstall whichever you want. So your laptop is probably safe.
It looks like EOMA68-A20 is not going to just work. I don't want it to end like my laptop.
Will the EOMA68-A20 boot from USB to run an installer?
-- hendrik
On Thu, Mar 9, 2017 at 2:37 PM, Hendrik Boom hendrik@topoi.pooq.com wrote:
Will the EOMA68-A20 boot from USB to run an installer?
right. it's complicated (down to the extremely small PCB size and the pinouts of the A20). feel free to skip right to the very last sentence. in between this is for technical readers.
first thing: the A20 is an "unbrickable" processor. you simply cannot disable MMC0 from being "first boot" option. period. you cannot put it into any kind of "DRM" or other "treacherous" mode. there are no e-fuses available to blow that would disable this "unbrickable" functionality.
please ensure that you have accepted - in its entirety - the prior paragraph before proceeding to read further.
so - bear with me.
(1) MMC3's pins come out of the A20 on the LEFT side. (2) MMC0's pins come out of the A20 on the RIGHT side. (3) NAND Flash comes out of the A20 at the BOTTOM set of pins (4) RGB/TTL also on the LEFT. (5) DDR3 comes out of the TOP. (6) HDMI comes out just around the bottom-right corner. (7) the PCMCIA connector is on the LEFT edge of the PCB (8) the HDMI, USB-OTG and Micro-SD are on the RIGHT.
with all that so far? now, here's the layout "restrictions":
(1) there are only 3 signal layers: TOP, BOTTOM and LAYER3. (2) it is NOT PERMITTED to cross anything over, in or between the DDR3 layout area, nor in the actual DDR area. (3) it is NOT PERMITTED to cross anything through the cental "power" area of the A20, because that would interfere with grounding of the central part of the SoC. (4) the amount of space ABOVE the DDR3 area is 83 mils (2mm) (5) the amount of space below the bottom edge of the A20 to the bottom of the PCB is also a scant 3mm and it's taken up on the right side by HDMI (and by the 24mhz XTAL) and on the left by NAND flash tracks.
all of that is as background to the following:
with NAND crossing over from the RIGHT side of the A20 to the LEFT, and with RGB/TTL going out from LEFT to the PCMCIA connector, there is simply NO WAY to route the MMC3 from the RIGHT side of the A20 over to the LEFT side where it could connect to the PCMCIA connector, and at the same time route MMC0 which is on the LEFT side over to the RIGHT of the PCB where the MicroSD card is.
it is also not possible to go DOWN from the A20 with either MMC0 or MMC3 to try to route AROUND the A20 (and DDR3) along the very top edges because:
* HDMI takes up the entire Layer3 and BOTTOM layer routing. * the 24mhz XTAL is in the way as well (blocking TOP layer) * various GPIOs have had to go along the top (spare) 2mm edge of the PCB, around the top of the DDR3.
basically there *is* no option to route MMC3 to the PCMCIA connector and MMC0 to the Micro-SD card, where it would be really really preferable to do that, because then the Card would be "self-unbrickable", i.e. you could put in USB power and a MicroSD card and boot (recover) straight away.
instead, because MMC3 (right) has had to be routed to MicroSD (right) and MMC0 (left) to the PCMCIA (left), if you want to recover an EOMA68-A20 Card using the "unbrickable" MMC0, you *need* a Housing of some description.
the first such Housing which will provide the "lowest cost, simplest" unbrickable option combo will be the Micro-Desktop. you *can* if you want make one yourself, from the breakout board and associated components.
now, to mitigate this scenario what i will be doing is (actually i'm forced into this option due to the NAND ICs slowly not being available any more), put in a very small (128mb) SLC NAND IC, on which will be installed boot0 and u-boot etc.
by default (and i already have this set up so it's going to the default... *by* default...) the u-boot on the SLC NAND will be set up to look for an OS on MMC3 *not* MMC0.
being SLC NAND it will not be corrupted by read operations, so as long as you don't try to screw with it by trying to write to it, it will be just as stable as the "unbrickable" MMC0, indefinitely.
summary: if you _do_ screw it up however, the "normal" option (MMC0 being on the Card) simply isn't available, instead you'll need to boot from a Housing that has an SD Card, such as the Micro-Desktop.
l.
2017-03-09 15:37 GMT+01:00 Hendrik Boom hendrik@topoi.pooq.com:
On Sun, 12 Feb 2017 19:35:00 +0000, Lyberta wrote:
Eric Duhamel:
Depending on what you mean by "newbies", I don't think they would know if they want any particular image of GNU/Linux except the one that was designed to run on the product by the maker of the product. After all, they don't know the nooks and crannies of GNU/Linux and would just want to receive a product that works.
OK, I'm a software developer, not a system administrator. I have no idea what 90% of packages installed on my system do. I only use terminal to upload my code to the Git repository. But I need g++ 6.
Debian Jessie has g++ 4.9.2 which is extremely old and none of my software will compile there. When I pledged for Debian card, I expected stock Debian with maybe a few custom packages which would be explicitly marked as such. And the first thing I'd do is to upgrade to Testing which as of writing this has g++ 6.3.0.
Now I'm told that issuing "apt-get dist-upgrade" is taking responsibility, etc. So I'm stuck with old and unusable frankendistro and on my own if I want to make it work.
I've chosen GNU/Linux because of its freedom and I've chosen Debian because it doesn't have proprietary software in main. I have no idea of what is going under the hood. I don't care what init system I run as long as it is free software and it boots my PC.
A couple of years ago I needed to buy a laptop. I've looked for one which doesn't come with Windows and I've found one with Ubuntu. Now, I really hate Ubuntu and the first thing I've done was to install Debian in dual boot. For some reason, Debian couldn't power off my laptop so I removed it and I'm still stuck with Ubuntu.
I know what you're going to say: "You should've looked for the solution on the Internet.". Well, sometimes I don't have time and am scared of bricking my hardware. I want things to "just work".
If you can boot your machine from USB, you won't brick it by changing the OS, because you can boot a Debian or Ubuntu installer and reinstall whichever you want. So your laptop is probably safe.
It looks like EOMA68-A20 is not going to just work. I don't want it to end like my laptop.
Will the EOMA68-A20 boot from USB to run an installer?
Not unless it's build.
Also most installers have limited support for non x86 devices. Intel platforms provide a somewhat "standard" way for booting external media. For ARM etc. there is no such thing.
For Intel you have the BIOS. For ARM you usually have u-boot.
When a A20 is powered. It reads from an internal ROM. This contains a minimal init program for loading a program and inits SRAM. This ROM is what Luke is fighting with because it cannot read from currently produced NAND Flash types. It searches in NAND and SD for a "magic" token and loads the data following that into internal RAM (sram) and than tries to run what's loaded. What's loaded is usually a minimal u-boot (SPL/boot0) which inits external RAM etc and loads another program, usally a full u-boot (mainline or boot1), from whatever location is programmed into u-boot SPL. That stage init's other peripherals and then loads the OS, usually linux from whatever location you specify.
boot0/boot1 are u-boot implementations by Allwinner. Mainline u-boot has support for A20. So no need for a butched version from AW.
u-boot does not a have a standard for searching locations and booting from those.
Since there is no standard, generic installers cannot work and thus installers must by tailored.
So for EOMA we must set/invent a standard which Installers can use regardless the SoC type used in the EOMA card.
Other ARM device vendor usually build their own installers. But that's a quick and dirty solution only only last as long as the vendor invests in that method.
On A20 et.al. there is the FEL mode. Which is initialized by that first loaded program. In this FEL mode you can push an image over USB-OTG. But that requires an active USB host not just an USB thumb drive.
http://linux-sunxi.org/FEL/USBBoot#Booting_U-Boot_over_USB http://linux-sunxi.org/BROM
So the question is are "we" going to build such a thing?
-- hendrik
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
On Sat, Feb 11, 2017 at 12:39:46PM +0000, Luke Kenneth Casson Leighton wrote:
this is a decision that is easily justifiable based on the fact that it's going to have to be distributed with the sunxi 3.4 kernel as that's the only one which supports the full hardware.
That's a proper technical argument, indeed. Anybody here managed to get the Jessie systemd running on the 3.4 sunxi kernel?
Also: I figure that currently mainline kernel support fo those is still not good enough for the image. But what are the chances for me to later have it running with mainline kernel and get a decent hardware support?
--- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Sun, Feb 12, 2017 at 9:23 AM, Tzafrir Cohen tzafrir@cohens.org.il wrote:
On Sat, Feb 11, 2017 at 12:39:46PM +0000, Luke Kenneth Casson Leighton wrote:
this is a decision that is easily justifiable based on the fact that it's going to have to be distributed with the sunxi 3.4 kernel as that's the only one which supports the full hardware.
That's a proper technical argument, indeed. Anybody here managed to get the Jessie systemd running on the 3.4 sunxi kernel?
Also: I figure that currently mainline kernel support for those is still not good enough for the image. But what are the chances for me to later have it running with mainline kernel and get a decent hardware support?
you'll have to hunt for the patch that was somewhere around 4.7rc0 to rc4 which causes the a20 cards to go unstable and crash arbitrarily within 30 to 200 seconds. i compiled about 100 versions of the linux kernel source trying to track down exactly where it was, but the above was as far as i could get. i had to stop as it was simply absorbing too much time right during the middle of the campaign.
i documented this on the updates.
[side-note directly to you tzafrir: btw i assume you'll be reading this, and that you've seen my response about libselinux1. i'm mentioning it because i have a vague recollection of answering your question multiple times, and each time you raising the exact same question demonstrating that you haven't seen my response. acknowledgement greatly appreciated so it doesn't keep on happening].
there appeared to be some significant chances to devicetree around that time and the cubieboard2 (which i was using as the basis for testing) was not kept up-to-date around that time.
once that bug which was introduced around that time has been found and fixed use of mainline kernel support should be absolutely fine...
but bear in mind that there is still a library needed to be written (which must go into u-boot as well) which reads the EOMA68 EEPROM at address 0x51 and loads the required "overlay" (devicetree fragment).
i just mention this just in case you were considering upstreaming an arbitrary devicetree file for the EOMA68-A20 - please don't, not without consulting me first.
l.
On Sat, 11 Feb 2017 12:39:46 +0000, Luke Kenneth Casson Leighton wrote:
ok, so, apologies for not responding for 2 days: the latest cold, which is back again a fourth time in as many weeks, is leaving me exhausted. again.
tzafrir: i've mentioned this a number of times, and am happy to mention it again, as you appear to have missed it. the key difference is massive scope-creep. look at how the NSA developed libselinux1 and associated infrastructure. they got a university involved to develop the FLASK model. they set out a design strategy, they set out what they were going to do, then they did it. whilst almost everything else that the NSA does may be questionable, steven smalley is clearly a smart guy and knows what he's doing.
by contrast the development of systemd has become a critical single-point of failure for a massive number of distros, where its developers are clearly and pathologically not taking responsibility for the consequences of their *technically-driven* decisions, and are continuing to develop their software without wider consultation.
so, i read what everyone wrote: i think the simplest thing to do is to just go with the image that i have been working with and testing over the past two years. it's using xfce4 (gnome is too heavy). i know it works, and i simply don't have the time - or importantly the energy - to create a new image, *especially* based on people's comments and reactions that they'd be deeply unhappy with it not being a "stock image", even if all i did was make it boot sysvinit instead by default. those comments *alone* immediately terminate all and any possibility that i can provide debian/jessie in a 100% ethical way.
Do the same you did with Debian, only use Devuan aand the Devuan installer. It will likely just work, and you won;t need to expunge systemd.
Sell it as a Devuan system.
You can still make a stock Debian available if your customers demand it. But by providing Devuan you won't be pushing systemd on those that don't want it. Your conscience can be somewhat clearer.
on receipt of the cards, anybody who wants to will be free and entirely at liberty to do "apt-get dist-upgrade"
That's one of the ways of installing Devuan for those that don;t want to set up fro scratch. Start with a running Debian system, Jessie or the one before that. Replace the lines in /etc/apt/sources.list by lines referring to Devuan's repository. DO an update, then a dist-upgrade.
-- hendrik
On March 7, 2017 5:20:50 PM PST, Hendrik Boom hendrik@topoi.pooq.com wrote:
On Sat, 11 Feb 2017 12:39:46 +0000, Luke Kenneth Casson Leighton wrote:
just go with the image that i have been working with and testing over the past two years. it's using xfce4 (gnome is too heavy). i know
it
works, and i simply don't have the time - or importantly the energy -
to
create a new image, *especially* based on people's comments and reactions that they'd be deeply unhappy with it not being a "stock image", even if all i did was make it boot sysvinit instead by
default.
those comments *alone* immediately terminate all and any possibility that i can provide debian/jessie in a 100% ethical way.
Do the same you did with Debian, only use Devuan aand the Devuan installer. It will likely just work, and you won;t need to expunge systemd.
Sell it as a Devuan system.
You can still make a stock Debian available if your customers demand it. But by providing Devuan you won't be pushing systemd on those that don't want it. Your conscience can be somewhat clearer.
I agree with this approach. Trying to single-handedly remove systemd from an otherwise maintained distro is targeting a symptom without mitigating the cause, and could break things that people expect to work. Devuan is a *great* solution for anyone wanting to get Debian without systemd, and there's no need to duplicate effort.
However, it is IMO too late to pivot to this solution as the cards have been tested on a specific image and changing that would invalidate years of costly testing. Plus, the bigger problem we have is that these cards can't be updated to current Debian *at all* because of a kernel bug.
That is why this is a somewhat limited run AFAIK and it's worth it to note Luke doesn't and can't have customers according to what he says. I find it likely the eventual distributor of mass-market cards will simply sell Debian as-is.
-- Eric Duhamel http://www.noxbanners.net/
OK, guys, so I have a smartphone that I deeply regret buying. I've installed Debian on it in chroot using Lil' Debi. A week ago I decided to try to make it useful and trying to update to Debian Testing. And when run "apt-get dist-upgrade", it said that glibc requires Linux kernel at least 3.2. My phone has 3.0.8 kernel.
And now I have totally useless piece of metal that can't even run modern glibc. I know I can install QEMU or something on it but how long QEMU will support old versions of packages installed on my phone?
And now I hear that A20 card will be shipped with kernel 3.4. Sure, it's not lower than 3.2 but how long will glibc support 3.4? What about other packages?
You say that EOMA68 is about reducing waste but the lack of support for mainline kernel means you are selling waste in some people's opinion.
I'm sorry, this is a perspective of a dumb customer. I have no idea how much work is needed for mainlining. I just want to say that a lot of people who don't follow this mailing list will be upset to see old kernel.
And now I hear that A20 card will be shipped with kernel 3.4. Sure, it's not lower than 3.2 but how long will glibc support 3.4? What about other packages?
Actually, mainline support for A20 isn't that bad, depending on your specific use case. Check http://linux-sunxi.org/Linux_mainlining_effort. It's fairly good for headless servers, less so for tablets and desktops. And there is (slow) progress, so its future isn't too bleak.
Stefan
On 03/07/17 21:33, Stefan Monnier wrote:
And now I hear that A20 card will be shipped with kernel 3.4. Sure, it's not lower than 3.2 but how long will glibc support 3.4? What about other packages?
Actually, mainline support for A20 isn't that bad, depending on your specific use case. Check http://linux-sunxi.org/Linux_mainlining_effort. It's fairly good for headless servers, less so for tablets and desktops. And there is (slow) progress, so its future isn't too bleak.
Stefan
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
I run Armbian with 4.9.12-sunxi on my Banana Pi which also uses an A20.
But of course there's no Mali anything for that kernel.
BTW there's some work on a GPL'd Mali 450 driver which is, I understand, the same as the Mali 400 in the A20, just with twice the shaders.
http://www.cnx-software.com/2017/03/06/mainline-linux-on-64-bit-arm-amlogic-...
On Wed, Mar 8, 2017 at 5:33 AM, Stefan Monnier monnier@iro.umontreal.ca wrote:
And now I hear that A20 card will be shipped with kernel 3.4. Sure, it's not lower than 3.2 but how long will glibc support 3.4? What about other packages?
Actually, mainline support for A20 isn't that bad, depending on your specific use case. Check http://linux-sunxi.org/Linux_mainlining_effort. It's fairly good for headless servers, less so for tablets and desktops. And there is (slow) progress, so its future isn't too bleak.
yeahhh, it's the lack of desktop support that's the main concern. i'd like to be able to say that it will be possible to fund the mainlining support but it isn't, right now.
also, as i mentioned earlier, the A20's running out of time. i'm going to have to ship with only an older toshipba 1MB or even just a 128kb NAND flash because it's getting harder and harder to find so-called "legacy" NAND flash ICs in TSSOP packaging that the A20's brain-dead boot ROM recognises.
so now you may gain some understanding why the RK3288 board is so important.
(a) it's got full mainline support (b) it's just as libre as the A20 Card (c) it has mass-volume support (d) it's had a chromebook and more made around it (e) it's been around long enough to have all the problems ironed out (f) the technical reference manuals are now online (g) it's possible to get Reference Designs for $USD 25.
that DOES NOT mean that the A20 board is not important: it is critical at this early phase.
l.
On Wed, Mar 8, 2017 at 5:02 AM, Lyberta lyberta@lyberta.net wrote:
You say that EOMA68 is about reducing waste but the lack of support for mainline kernel means you are selling waste in some people's opinion.
then people had better pick up the pace and fix that situation, then. that's why i offered to send out some early systems to people.
i can't do everything - i'm barely coping with the hardware design as it is.
l.
There is a bug that prevents it from booting on newer kernels.
On Wed, Mar 8, 2017 at 7:02 AM, Lyberta lyberta@lyberta.net wrote:
OK, guys, so I have a smartphone that I deeply regret buying. I've installed Debian on it in chroot using Lil' Debi. A week ago I decided to try to make it useful and trying to update to Debian Testing. And when run "apt-get dist-upgrade", it said that glibc requires Linux kernel at least 3.2. My phone has 3.0.8 kernel.
And now I have totally useless piece of metal that can't even run modern glibc. I know I can install QEMU or something on it but how long QEMU will support old versions of packages installed on my phone?
And now I hear that A20 card will be shipped with kernel 3.4. Sure, it's not lower than 3.2 but how long will glibc support 3.4? What about other packages?
You say that EOMA68 is about reducing waste but the lack of support for mainline kernel means you are selling waste in some people's opinion.
I'm sorry, this is a perspective of a dumb customer. I have no idea how much work is needed for mainlining. I just want to say that a lot of people who don't follow this mailing list will be upset to see old kernel.
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
arm-netbook@lists.phcomp.co.uk
-
Adam Van Ymeren -
Albert ARIBAUD -
Allan Mwenda -
Benson Mitchell -
Bill Kontos -
Boris Barbour -
Christopher Havel -
Elena ``of Valhalla'' -
Eric Duhamel -
Hendrik Boom -
Isaac David -
Jonathan Frederickson -
jtd -
Julie Marchant -
Luke Kenneth Casson Leighton -
Lyberta -
Michael Howard -
mike.valk＠gmail.com -
Paul Boddie -
pelzflorian (Florian Pelz) -
Philip Hands -
Ralf-Peter Rohbeck -
Russell Hyer -
Stefan Monnier -
Tzafrir Cohen -
Vincent Legoll -
Wolfgang Romey -
zap