Re: [Cvs-dev] CVS maintenance question

[Thorsten Glaser]

Hello everyone! (Assuming this mail goes through.)

Assaf Gordon dixit:

> Second,
> If you have any recommendation as to the best way to proceed, any
> critical patches you think should be applied - all these and any
> other comments are welcomed.
> For example, why did Centos/Redhat/Fedora choose 1.11.23, while
> Gentoo - 1.12.12, and Debian/Ubuntu used 1.12.13 ?

I’m the MirBSD developer, and the maintainer of CVS in Debian.
I’ve ensured to use the same code for both operating systems,
and as such I updated to the last available upstream version,
and then started to do bugfixing and development of new features
(like “cvs suck”) on my own, in a whole-tree approach (i.e. no
series of patches, but working on the source as a whole, inside
MirBSD).

As such, I probably lost some portability (assuming arc4random(),
for example, which is easily provided by libbsd on GNU systems,
such as Debian, though).

> For savannah we need some features that were committed after 1.12.13,
> so we can't even use the packaged version.

Interesting. I haven’t looked at what was committed upstream
after the last release, only recently seen there was (until
2008), and I’d have to check and see what of it is actually
complete and usable.

Can you list which exact post-1.12.13 features/changes you need?

I’m not yet sure if I can get them into the next Debian stable,
as the freeze process has already started, though.

> If any one of you is interested in stepping up to be the chief
> upstream maintainer of cvs - that would be welcomed.

If nobody else wants, I still feel for GNU CVS 1.12.x enough to
be willing to maintain it as new upstream (I had big hopes for
OpenCVS only to be utterly disappointed) despite its licence and…
well, dusty codebase and environment (gnulib is the cause of much
agony). However, I have never been involved with GNU CVS upstream
in any way, and I’m not very experienced in the particulars of how
it works internally (although getting better), so…

> Some current CVS maintainers expressed willingness to help to any new
> maintainer.

… this would definitely be a big help.

> Perhaps this could usher a new long-due release.

I’d probably start getting whatever is currently in Debian and
MirBSD into portable shape, which *requires* external review as
I don’t deal with that much, and putting that out, and as a second
step, merging, peu à peu, whatever was done in GNU CVS post-1.12.13
(as it was never released, I’m not considering just merging its
HEAD at the moment, unless the previous maintainers state it’s
good enough to not introduce any regressions or showstoppers),
probably in the order in which people need it.

I’d also wish to open discussion on some issues, mostly reported
by users in the Debian BTS over the years, and could use help from
people with insight into how VCS in general and CVS/RCS specifically
work.

I’m not all that easy to work with, but no Theo de Raadt either ;)

I don’t have as much time for my OSS work as I used to have, these
days, but considering CVS is just as slow-moving a target as MirBSD
I could probably manage. I would seriously cut down on infrastructure
though (e.g. one mailing list is enough, no Forge needed, etc).

bye,
//mirabilos
-- 
FWIW, I'm quite impressed with mksh interactively. I thought it was much
*much* more bare bones. But it turns out it beats the living hell out of
ksh93 in that respect. I'd even consider it for my daily use if I hadn't
wasted half my life on my zsh setup. :-) -- Frank Terbeck in #!/bin/mksh