On Tue, Feb 18, 2014 at 1:32 PM, Neil Bothwick <n...@digimed.co.uk> wrote: > On Tue, 18 Feb 2014 22:07:12 +0400, Andrew Savchenko wrote: > >> > Then write. Just be aware that to write a systemd profile, you need to >> > use systemd. >> >> Or to create a non-systemd profile :) > > We already have many of those, because systemd is not the default. Part > of making it the default, if that decision is ever taken, would be to > modify the current profiles to support systemd, at which point the old > versions would become the non-systemd profiles. > > Yes, it does take systemd users/devs to create a systemd profile, but > they are the one that will want to use it anyway. The rest already have > what they want. > > This is the way things have moved with the GNOME and KDE profiles, expect > others to follow suit.
After seeing Andreas K. Huettel response in the other thread[1], I think it's fine even with a systemd profile. It just sets the systemd USE flags, mask genkernel (and dracut is so much better, IMO), mask some USE flags (static-flags for udev, cryptsetup, and lvm2; static for dmraid; and consolekit in general). It's really simple; putting that on a profile or doing by hand (which I do in my no-GNOME servers) is the same to me. > I'm still uncomfortable with the pervasiveness of systemd, although Canek > does put forward persuasive arguments, through a mixture of expertise and > remaining calm. Thanks Neil. > So GNOME want to use logind, which may well be superior > to ConsoleKit, but why should that require a change of init system? Well, the logind dbus interface is available for anyone to implement independently from systemd[2]. Ubuntu is trying to do that. It's just that the systemd developers saw that using the features of systemd, doing user session management was really easy, and they did. Those systemd features are not gratuitous; that's why Ubuntu is having trouble doing an independent replacement. > A login daemon should be started by the init system, not be an integral > part of it. What happens when logind no longer fulfils developers needs, > as is the case with ConsoleKit now, how can it be replaced with an > improved service when it is so closely tied to the init system. Well, if that happens then they will use the support for the improved service and logind will die like HAL or devfs. The thing is that logind exists now, it solves real problems, and people are using it because of that. If someone else writes something better, I'm sure they will use that instead. I don't see the point on worrying about what could happen when dozens of technologies have already been tried in Linux; some strive, and some die. Apparently, Upstart will die; it was a waste then when RedHat choose it for RHEL 6 (or 5, I don't know, never used it), or that Ubuntu used it? No, lessons were learned from it. And from devfs, and OSS, and HAL. That's how free software evolves. It only needs people willing and able to write and maintain new cool software. Regards. [1] http://article.gmane.org/gmane.linux.gentoo.user/272668 [2] http://www.freedesktop.org/wiki/Software/systemd/logind/ -- Canek Peláez Valdés Posgrado en Ciencia e Ingeniería de la Computación Universidad Nacional Autónoma de México
