On 12/7/16, Philip Hands phil@hands.com wrote:
Luke Kenneth Casson Leighton lkcl@lkcl.net writes:
On 12/5/16, Julie Marchant onpon4@riseup.net wrote:
by contrast: fvwm2 is an 8 *megabyte* install size. gnome is... what... several hundred megabytes? latest versions force you to use wayland? and systemd?? fuck that!! absolutely no way i'm tolerating that.
GNOME does not force you to use Wayland. I don't know where you got this idea from. Wayland is still supported experimentally (X is used by default, Wayland support is quite buggy) last time I checked. As for systemd, GNOME requires logind, but not the entire systemd package.
that requires libsystemd, which i refuse to have on any machine that i am managing.
I really think that you should draw a line between libsystemd0 and systemd itself (whether running as init or not).
One of the functions that libsystemd provides is the ability to check whether systemd is available, so if you want there to be a vibrant ecosystem of packages that do not require systemd, then it's probably worth encouraging people to write programs that check if systemd is available, and then behave sensibly if it is not.
The alternative forces the systemd refuseniks to fork every package that might find any (even optional) use for systemd services, which a) there is not sufficient manpower for, and b) removes any pressure for the upstream to put effort into accommodating your needs, so they don't bother to maintain/add the conditional code.
not sure i fully understand what you're saying, but i'm aware that devuan is supporting a huge alternative range of init systems: the only one they *don't* support is... systemd!
now, it may surprise you to learn that i've spoken to them and pointed out to them that if they want to not appear to be hypocritical (i.e. directly at odds with their stated goal of being "inclusive"), they really, *really* need to include systemd as one of the options.
however, because they've gone the "reacting against" polarisation route, which is just as equally bad as the forced-adoption route done by pretty much everyone else, there's still a lot of bad blood that needs to be healed first before anything like that can even *remotely* be considered.
That said, I don't have a lot of time for Gnome either, but that might be because a) I prefer Xmonad so I'm not their target audience,
i love the description you gave me a few years ago of what xmonad is capable of in such a ridiculously small amount of code... :)
and b) I run Debian, and we're making life difficult for Gnome maintainers by continuing to ensure that using systemd as init is optional (unlike most other distros),
good! about time someone stood up to the railroading but without the polarising perspective taken by devuan and other non-systemd distros!
and also constraining systemd when it is running as init to be backwards compatible with sysvinit in various ways, which means that there are things that Gnome can safely assume on the likes of Fedora which might not be true on a particular Debian install, so I guess Gnome on Debian has interesting little bugs that don't appear elsewhere.
well, that's actually really really good, because it means that the *BSDs don't get railroaded (through the unbelievably arrogant expectation of the systemd team that the *BSDs, *particularly* the high security-conscious ones, will simply roll over, take it up the backside and do the systemd team's bidding by adopting systemd or its interfaces or libraries).
If everyone that doesn't like systemd runs screaming away from Debian, shouting about libsystemd0, and doesn't bother to report bugs where our ambition to support other inits falls short, then they just ensure that the future they fear comes to pass.
y'know... the current hypothesis i'm floating in my head is that the full-time paid-up software libre projects are running at such a faster pace than the volunteer-driven ones that the full-time paid-up developers completely swamp and overwhelm the volunteer-driven ones.
also because they're on a different kind of motivation ("must get results otherwise my boss will fire me or not give me a good performance review") they feel *obliged* to not interact with the volunteer-driven teams.
how can i propose this hypothesis? well... over many many years, all of the projects that people have had real problems / issues with in the wider software libre community have been ones that are full-time funded.
so the key problem is that there's no real respect or inter-project communication... and, crucially, *no real reason for them to*. each project is "head-down getting on with it". as in: the actual task of developing *code* doesn't actually *need* inter-project communication or coordination. each project is doing their own thing, and doing it well.
still thinking about this and keeping an eye on it... but gotta go.
thank you for your insights, phil. really appreciated.
l.