Re: On our package plan
[2019-05-14 13:32] Jonas Smedegaard <jonas@jones.dk>
> Quoting Dmitry Bogatov (2019-05-09 09:35:13)
> > We have pandoc[1], haskell-intern, haskell-swish that are maintained
> > separately from DHG_packages, we have cborg-json, which don't even
> > have Vcs-Git field, probably more packages in unexpected places. My
> > point is that we need more unification.
> >
> > [1] Jonas, I remember that you dislike team mainenance, but let me
> > try to convince you.
>
> For the record I don't "dislike team maintenance" but a) like each code
> project tracked separately (i.e. a "pandoc" git not "all-haskell" git)
> and b) don't understand the meta machinery and would prefer to not need
> to learn it (i.e. that others in the team deal with that).
I understand you[1], but problem is that haskell packages are very, very
inter-connected. Essentially, all haskell libraries are parts of very
fragile whole (run-dependencies are hash-based). You can't just upgrade,
for example, libghc-pandoc-dev, without making sure that rest of
packages can be re-built with new version, or cabal-hell will happen.
In a way, I envy Python folks -- random developers keep packaging random
libraries, get them uploaded by random sponsors and things do not shatter
into hundreds of pieces.
As I am writing this, for time being I imported missing patches
semi-manually into package plan, but encountered unexpected hurdle --
cabal(1) crashes. I do not know how to fix it yet, and I do not know how
to proceed futher with packaging libraries for ghc-8.6.
If anyone is interested, I work on branch wip/kaction/repack-index-2 in
"package-plan" repository, and crash of cabal(1) reproduces on Gitlab
CI[2].
[1] Severals years ago I wanted to have blog-literately in Archives, and
did not want to go into package planning. Sigh.
[2] https://salsa.debian.org/haskell-team/package-plan/-/jobs/176658
--
Note, that I send and fetch email in batch, once every 24 hours.
If matter is urgent, try https://t.me/kaction
--
Reply to: