Hello,
On Sat, Jun 17, 2023 at 08:55:24AM -0400, pa...@quillandmouse.com wrote:
> Debian's capable of deciding that systemd is installed by default,
> etc, even if such decisions occur after monumental
> discussions/arguments.
It took literally years and is still hotly revisited from time to
time by various aggrieved parties on both sides.
It's highly unlikely you will get enough people to care about there
being one true package management frontend to do the same thing for
that.
If you think I'm wrong, good news: Debian does have a procedure for
making such decisions - the General Resolution! So just find a Debian
Developer who's willing to call a GR on this subject and five others
who are willing to second their call, and off we go. It will be
interesting if that happens, and I will only hate it if it causes
"apt" to change how it works in any way whatsoever.
However, another of those tricky core principles of Debian is that a
GR can't force volunteers to do work, so even if someone did succeed
in passing a GR that said that there would be one way of package
management, or that Synaptic has to have X, Y and Z done to it (I
don't know anything about Synaptic, so can't give any concrete
examples of missing features), it STILL won't happen without
agreement by the upstream developer(s) of those packages.
Debian has passed GRs before that tried to make people do things,
and found that no one wanted to do them, so despite there allegedly
being project consensus that "Something Must Be Done", the thing did
not get done after all.
> I'm not suggesting there should be one package manager to rule them
> all. But when the various tools resolve dependencies in different ways,
> we clearly have a problem.
I guarantee you that there are people reading this who enjoy
that their package management frontend of choice resolves things the
way it does and would be bitterly upset if that were to change. But
even they are not the ones who decide. The authors of the software
decide. Once the upstream authors and the Debian package maintainers
(these two groups may or may not overlap) have decided something,
there is basically no authority within Debian that can overrule
them.
This is currently happening with dpkg by the way - its author
objects to merged-/usr.
This is a fundamental aspect of the Debian project and butting your
head against it will leave you in the same place you started at, but
with a hurt head.
> A problem which could be resolved, at least in theory, by a
> concerted effort on the part of Debian leadership.
A read into the history of merged-/usr may be enlightening. Right
now we have a project-wide consensus that it should be done, but the
dpkg author fundamentally disagrees with it [being done in dpkg], so
the dpkg patches that Ubuntu and other derivatives have carried for
literally years are not in Debian. External means have had to be
implemented to convert existing systems to merged-/usr because the
dpkg author won't co-operate with the process and no one can force
them to do so.
So if Debian, with a project-wide consensus, can't carry out a major
change that has already been done in its biggest derivative years
ago, do you really think that it would be any easier to start
mandating how package management frontends work, over the wishes of
multiple upstream authors?
> Getting someone to implement the solution would require an eager
> volunteer.
You understand that the eager volunteers that Debian has already are
what resulted in there being multiple package management frontends,
right?
And that there isn't just one of them because there is no authority
within Debian to tell the other n-1 of them that their ideas are
unwelcome?
As mentioned, a GR could do it in theory, but you will never get the
support because of the ethos of Debian. It's just not that kind of
project.
> I'm not here to snipe at Debian.
You are asking a lot of "Why can't Debian behave in ways it never
has and would be a dramatic departure for it to start doing now?"
style questions though.
> As regards Synaptic, based on the discussion here, I'm guessing that
> whoever is maintaining it has no particular desire to craft a Wayland
> version, and that's why it hasn't been done. Still, Debian leadership
> could put out a call for volunteers to work on this. It *could* be
> done.
Again, I don't use Synaptic so am not aware of what functionality is
missing exactly, but yes, in general if there are volunteers and the
upstream authors are amenable then of course it's just a simple matter of
programming! You won't get a mandate from Debian leadership because
Debian leadership doesn't tell volunteers what to work on.
That it hasn't happened already suggests that there aren't any
sufficiently motivated and skilled volunteers (or that there are and
the author is blocking it for some reason - haven't looked).
I mean, is there even a bug report for this and did the maintainers
engage with it? That is the proper way to proceed
