[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: ghc now almost ready for migration



Hi Mehdi,

thanks for the investigations.

Am Montag, den 06.06.2011, 18:16 +0200 schrieb Mehdi Dogguy:
> On 04/06/2011 13:26, Joachim Breitner wrote:
> > 
> > with the upload of haskell-debian we have finally migrated (or removed)
> > all packages to ghc instead of ghc6. Once -debian has waited for its 10
> > days, I expect that the migration is possible, but of course there could
> > be unforseen difficulties, e.g. an arches where some package did not
> > build.
> > 
> > Can you already put the hints in place and find out whether the
> > migration will work in 10 days?
> > 
> 
> So, I've tried to see what's needed to be able to migrate ghc to testing.
> I've found the following issues/blockers:
> - pandoc out-of-date on several architectures. It seems that it needs
> highlighting-kate to build which is missing there. fwiw, gitit needs pandoc.
> - haskell-dummy has to migrate but it needs (at least) pandoc.

highlighting-kate is one of the big beasts that tend to fail on weaker
arches; on some of them it works with some buildds and not others.
Should I
 * give back the package once or twice to find out if there is a strong
buildd or
 * should I request removal of the out-dated packages (and depending
packages) on these arches, to allow for the transition to happen?

> - haskell-hsql-mysql needs mysql-5.1… which is not going to migrate soonish.

Ah, hmm. Should we then postbone the haskell transition and keep
updating stuff (e.g. the just released new ghc)? Or will you force-hint
haskell-hsql-mysql with a subsequent binNMU in testing to fix the
package there? Or do you want to remove the package in testing (seems
like it, given the hint file)?

> - haskell-llvm is ood on kfreebsd-i386.

Strange issue. I tried a give-back, the build log is rather old and
maybe the problem disappeared.

> - haskell-leksah-server is ood on mips (needs haskell-haddock).

haskell-haddock fails with error 9, sounds like OOM killed or timed out
or something related. I’ll ask mips@buildd.debian.org if they have a
timeout or something.

> - haskell-pretty-show is ood on mips/mipsel (needs haskell-lexer).

OOM for haskell-lexer. I guess we have to remove it from these arches.



> If we try to migrate the whole of haskell stuff (without haskell-dummy),
> britney fails with the following:
> 
> endloop: 87+0: i-1:a-0:a-0:i-0:k-43:k-43:m-0:m-0:p-0:s-0:s-0
>     now: 134+0: i-33:a-3:a-1:i-0:k-46:k-46:m-1:m-1:p-1:s-1:s-1
>     * i386: configfile-doc, ftphs-doc, gitit, haskell-agda-doc,
> haskell-convertible-doc, haskell-cpphs-doc, haskell-edison-api-doc,
> haskell-edison-core-doc, haskell-haskelldb-doc, haskell-hdbc-doc,
> haskell-hdbc-odbc-doc, haskell-hdbc-postgresql-doc,
> haskell-hdbc-sqlite3-doc, haskell-hscurses-doc, haskell-hsql-doc,
> haskell-hsql-mysql-doc, haskell-hsql-odbc-doc,
> haskell-hsql-postgresql-doc, haskell-hsql-sqlite3-doc, haskell-http-doc,
> haskell-pcre-light-doc, haskell-regex-base-doc, haskell-regex-compat-doc,
> haskell-regex-posix-doc, haskell-src-exts-doc, haskell-uulib-doc,
> haskell-zlib-doc, ldap-haskell-doc, libghc6-gitit-dev, libghc6-pandoc-dev,
> magic-haskell-doc, missingh-doc
>     * amd64: gitit, libghc6-gitit-dev, libghc6-pandoc-dev
>     * armel: libghc6-pandoc-dev
>     * kfreebsd-amd64: gitit, libghc6-gitit-dev, libghc6-pandoc-dev
>     * kfreebsd-i386: gitit, libghc6-gitit-dev, libghc6-pandoc-dev
>     * mips: libghc6-pandoc-dev
>     * mipsel: libghc6-pandoc-dev
>     * powerpc: libghc6-pandoc-dev
>     * s390: libghc6-pandoc-dev
>     * sparc: libghc6-pandoc-dev
> 
> FWIW, the attached file haskell.txt has my britney hint. Remove
> haskell-dummy from the final list if you want to see the failure.
> Otherwise, it will say only: “failed: haskell-dummy”.

> FWIW, only gitit (in testing) build-depends on libghc6-pandoc-dev and
> nothing build-depends on libghc6-gitit-dev. In sid, libghc6-pandoc-dev and
> libghc6-gitit-dev are not used at all. So why not simply getting rid of
> them? I don't see the point of keeping them around.

gitit in unstable build-depends on libghc-*-dev, without any 6 in the
name – shouldn’t gitit just migrate along? Or maybe I’m not following
you here...

Greetings,
JOachim

-- 
Joachim "nomeata" Breitner
Debian Developer
  nomeata@debian.org | ICQ# 74513189 | GPG-Keyid: 4743206C
  JID: nomeata@joachim-breitner.de | http://people.debian.org/~nomeata

Attachment: signature.asc
Description: This is a digitally signed message part


Reply to: