Why are we based on Ubuntu and not Debian?
Looking at https://www.gnu.org/distros/common-distros.html it sounds Debian is far better than Ubuntu in case of respecting freedom. Furthermore Debian is a community based project whereas Ubuntu is commercial distribution owed by Canonical Ltd. So, I think we should prefer Debian as a base rather than Ubuntu.
Hello, glad to see you in Trisquel Forum! As far as I know, this issue
is often discussed already. Personally, I don't think changing base
would accelerate or enhance Trisquel development especially for now.
And let's suppose it happens, next time somebody could come and could
say similar thing "why are we based on Debian and not painly
GNU+Linux?". All in all, I really like you showing your commitment to
software freedom.
Cheers,
Ade
There have been already so many posts discussing this issue. I'm also negative toward Ubuntu as a base.
If you prefer Debian, you can use it anyway. The main freedom issue of Debian is the existence of the (unofficial) non-free repository which provides non-free firmware. But it's not enabled by default. Avoid to enable it and you should have a free/libre installation.
I would vote for basing off of Devuan. With Debian and Ubuntu you are locked into too many of RedHat's walled garden technologies going forward.
> I would vote for basing off of Devuan. With Debian and Ubuntu you are locked into too many of RedHat's walled garden technologies going forward.
I think Debian is a better choice than Devuan for an upstream distro, at least these days. Starting with Bullseye (and with Buster if you backport Bullseye's version of elogind) it is not very hard to base a systemd-free distro on Debian.
I have looked through Devuan's VCS, and most of their modifications don't actually improve SysVinit compatibility, but rather are Stalinesque removals of any reference to systemd ever existing. Doing this isn't harmful, but it's a waste of time that could be and have been spent improving SysVinit compatibility and keeping up with Debian releases. Contrast with AntiX, who unlike Devuan does not spend time removing references to systemd, but has never depended on libsystemd0 and does a much better job keeping up with Debian releases. Trisquel already has trouble keeping up with Ubuntu releases. Imagine if after a Debian release, Trisquel users had to wait for Devuan to release their downstream version, and *then* wait for Trisquel to release theirs.
Also, because Devuan does not (and does not have the resources to) maintain SysVinit support themselves, if Debian were ever to drop SysVinit support it would almost certainly mean the end of Devuan, so using Devuan as a base does not actually make you less dependent on Debian's decisions in the long run.
If you are just a user who just wants a systemd-free Debian system working out of the box, Devuan is a fine choice, but as the base for a distro I think Debian is a better option. On most of my machines I currently run my own customized version of Debian, following AntiX's approach for SysVinit compatibility.
Ubuntu on the other hand, is indeed a poor choice if you want to avoid using systemd. Whereas Debian simply does not prioritize SysVinit support, Ubuntu makes conscious decisions to make use of systemd's features in some of their own software.
I'm so happy to see you continue to dig into antiX's plumbing. Isn't it an absolutely wonderful conceptual approach?
I did not know that Devuan was wasting time removing those references. That's an interesting point, I'm going to need to explore further. As far as waiting on Devuan, they do offer up a testing and an unstable branch, so you aren't forced to wait the extra year for them to release their stable version. And I do think their community will continue to grow. Beta 3 of Beowulf is quite impressive, and should attract a lot of attention when they announce Beowulf stable version in the next few weeks.
> I'm so happy to see you continue to dig into antiX's plumbing. Isn't it an absolutely wonderful conceptual approach?
I was initially thrown off by the [1][nosystemd repository], which currently contains a single package, but used to contain a large number of packages that were no longer actually needed or used in antiX. The key is that starting with version 241 elogind is ABI compatible with libsystemd0, meaning that it works as a drop-in replacement with no need to recompile packages that have been compiled against libsystemd0. Unfortunately elogind 241 missed the window to make it into Debian buster, which has elogind 239, but antiX backports [a newer version][2]. Basically, these packages
* eudev
* elogind >=241
* [libpam-elogind-compat][3]
* sysv-rc
can replace systemd and leave Debian otherwise functional, barring the occasional missing init script, which [sysd2sysv][4] makes relatively easy to create from a systemd unit file.
> I did not know that Devuan was wasting time removing those references. That's an interesting point, I'm going to need to explore further.
Maybe my "wasting time" comment was unfair. I don't know how much time they actually spent on it or whether or not it's a bottleneck in development. [Here's a pretty typical example][5] though. Debian's lightdm package works fine with or without systemd. Devuan's changes just remove systemd support. This has no benefit for non-systemd users, which presumably includes all Devuan users.
[1] http://repo.antixlinux.com/buster/pool/nosystemd/
[2]: http://repo.antixlinux.com/buster/pool/main/e/elogind/
[3]: https://packages.debian.org/experimental/libpam-elogind-compat
[4]: http://www.trek.eu.org/devel/sysd2v/
[5]: https://git.devuan.org/devuan-packages/lightdm/commit/f57ed155705a5b41a1879057e2a8b50ae1d6394f
> Debian's lightdm package works fine with or without systemd. Devuan's changes just remove systemd support. This has no benefit for non-systemd users, which presumably includes all Devuan users.
That's a fair point. Devuan's devs appear to have a specific philosophy of not supporting systemd as a counterpoint to Debian. Antix's top dev is anticaptilalista, who has made public statements saying that Antix will never include systemd, but he seems more than happy to work hand-in-hand to help the MX devs develop their distro. MX offers both sysvinit and systemd, with the default being sysvinit, and the user getting the choice at boot time. So Antix would be harming MX which is downstream if it got rid of systemd support. (or maybe "downstream" is not the correct term, but Antix is the distro that gets released earlier, and MX uses Antix's code and releases a few months later).
There's always PureOS. Based on Debian testing and endorsed by the Free Software Foundation. Runs flawlessly on my Libreboot x200 and the software repos are nice and up to date.
/me hides before the anti-Purism assassins show up.
I've been installing GNU/Linux since the early 2000s, but I've never yet successfully installed PureOS. Also, it only releases 64-bit versions, so I can't even try it on my older laptop. I'm definitely not one of the Purism disestablishmentarians, but I don't really see the benefit of using PureOS over vanilla Debian (or Trisquel) if you're installing GNU/Linux yourself. I definitely wouldn't suggest trying to base another distro off PureOS.
> Looking at https://www.gnu.org/distros/common-distros.html it sounds Debian is far better than Ubuntu in case of respecting freedom.
This has been discussed over and over and over again, but I'll briefly explain. When it comes to the actual software in their repositories, which is what matters to a downstream distro trying to comply with the FSDG, Debian and Ubuntu are pretty much equally free. The difference between their distros is their attitude. Debian calls their non-free repositories "unofficial" and somewhat discourages their use, disabling them by default when users install Debian. Ubuntu does not make efforts to distance themselves from their non-free repositories and enables them by default after installation of Ubuntu. However, this is completely irrelevant to a downstream distro.
There are other reasons I personally prefer Debian over Ubuntu, but they don't have anything to do with the FSDG.
Yes, actually due to suffering from this issue https://trisquel.info/en/issues/23853 for more than 2 years, I've switched to PureOS just 2 days ago.
Additionally, if Trisquel still wants to support i686 architecture, then it must make some change starting version 10, since Ubuntu dropped i686 support.
> Additionally, if Trisquel still wants to support i686 architecture, then it must make some change starting version 10, since Ubuntu dropped i686 support.
There has been some discussion of this in recent meetings. Ubuntu still provides some 32-bit binary packages in order to support Wine and Steam, but not for the entire repository. Trisquel is soon to upgrade to a new development server, after which it may be feasible to compile all of the missing binaries ourselves, allowing Trisquel to continue to support i386.
Thank you very much for the reply. I still keep a few 32-bit computers, including two ThinkPad X40, currently used for GNU/Hurd testing.
Since I probably won't use said 32-bit computers to run latest packages (i.e., for non-critical tasks only), I may consider to install incoming Trisquel Mini on one of them. I really enjoy the desktop environment customization made by Trisquel.