Re: Request to fast track gitlab dependencies
> To be frank,
>
> The reason for that is that no one sane would recommend running software
> which changes at high frequency.
>
> I think the core of your problem is not that "gitlab is complex", as you
> sometimes claim on various Debian mailinglist, but more that "gitlab
> changes far too quickly for Debian".
In my opinion one of the biggest advantages of debian for the open
source movement is that it made many software packages provide Long Term
Support versions. Also a few base packages (the kernel, the build
system, some components that can be used on web servers) are maintained
in a way that really makes a rock-solid system.
But if one wants to use Debian as Desktop OS that interfaces with the
Real World one tends to run into problems:
- File formats, Web Protocols and APIs can change after the release of
Debian and can require big changes if one wants to fix them with a
simple patch.
- some packages during the LTS period change so much that finding the
part of the code that corresponds to a bugfix/ security fix is quite hard.
- and some programs, notably everything that has to do with KDE,
Gitlab, Google APIs or npm just follow a rolling release cycle.
We cannot tell the world that it doesn't want to work with all of these
things. And we cannot tell users that want to use them that they don't
want to use debian because - the stable base system is something many
users don't want to miss. And even if I personally use git versions for
most applications I use for productive work I want to build them on a
system that is guaranteed not to cause additional problems.
=> The proposal to provide several package sources:
- An ultra-stable base system
- Backports for the feature-hungry that originate in a tested release
- and something that is fast-paced enough for Rolling Release
applications but somehow follows rules that allow to provide a minimum
of stability
doesn't sound like a bad idea especially as there would be volunteers
for working on the details.
Kind regards,
Gunter.
Reply to: