Andrej Shadura writes ("Re: Consensus Call: Git Packaging Round 1"):
> I noticed some people [citation needed] think it is not important to
> preserve pristine upstream tarballs with the move to Git, and it's
> okay to regenerate them from a Git branch without trying to preserve
> checksums of the tarballs upstream has somehow generated.
I am one of these people. I have always been sceptical of the need to
preserve upstream pristine tarballs. I haven't been vocal about this
because no-one is forcing anyone to publish pristine tarballs. So in
any situation where the maintainer doesn't want to pay the costs of
preserving pristine upstream tarballs, the maintainer can simply not
do so.
That overall stance has a lot of social value for the project, because
it means we can all cooperate without having to have this debate. We
can save our energy for doing something more useful.
For myself I try to publish pristine upstream tarballs when it's
reasonably convenient, because I feel that this is a thing that some
people value, even though I myself think the value is very limited.
That's part of playing nice in a community like Debian.
If we are going to mandate something - or even, if we are going to
change our current stance (which seems to be that this is a "nice to
have"), then a discussion of the upsides and downsides - particularly,
with a practical focus - is necessary.
> I just had an impression this is not being considered and sort of
> assumed a consensus that we won't keep them in Git when we move to
> git-debpush workflows. Let's discuss it instead of have people
> assuming things potentially incorrect :)
Adoption of git-debpush will be optional, obviously. I think reasons
to actively encourage it over `dgit push-source' [1] are rather
limited, at least right now.
Also, in principle git-debpush and the tag2upload service could
support eg pristine-tar, although that would involve quite some
development work. If someone wants to lead that, I would be happy to
review patches, do supporting work, etc.
So certainly, speaking for myself, I don't intend to discourage or
impede anyone from publishing pristine tarballs, even though as I say
I think their value is very limited.
Sam Hartman writes ("Re: Consensus Call: Git Packaging Round 1"):
> That's on my list for discussion in round 3.
I have carefully avoided any discussion of the merits in my message,
and am happy to wait.
Ian.
[1] Both `dgit push-source' and `git debpush' publish your actual git
history on the dgit git server in a form useful for users, so I have
argued, and will continue to argue, that there are good reasons to
encourage maintainers to do one of these things.
--
Ian Jackson <ijack...@chiark.greenend.org.uk> These opinions are my own.
If I emailed you from an address @fyvzl.net or @evade.org.uk, that is
a private address which bypasses my fierce spamfilter.
