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

Re: utf8-string in ghc6



Hey,

On Wed, Feb 17, 2010 at 11:38:49AM -0200, Marco Túlio Gontijo e Silva wrote:
Hallo Joachim.

Excerpts from Joachim Breitner's message of Qua Fev 17 11:08:54 -0200 2010:
(...)
while creating a patch against ghc6 to run dh_haskell_provides I notice
that ghc6-6.12 ships utf8-string and time, which we currently also ship
as individual packages:
(...)
The time is up-to-date WRT hackage, utf8-string is slightly behind
(...)
Is there a point in maintaining individual packages for these, or should
we just drop them, as it is the case for Cabal etc.?

I think we should drop it unless we have a package that depends on the newer
version of utf8-string.  I've done a research about it:

$ for i in `grep-dctrl -FDepends -sPackage libghc6-utf8-string-dev /var/lib/apt/lists/localhost\:9999_debian_dists_sid_main_binary-amd64_Packages | sed 's/Package: libghc6-\(.*\)-dev/\1/; /Package/d'`; do echo ---$i; GET http://hackage.haskell.org/package/$i | grep -A 1 utf8-string || echo $i >> /tmp/error; done | less

Then I searched by hand the pacakges in /tmp/error.  The only problem I've
found is with the newer version of haskeline, which is not in Debian yet.  If
we need it, we'll need a separate package of utf8-string.  As it's only a
possibility, I'd say we should drop them, and reintroduce then if it's the
case latter.

Greetings.

Thanks for the analysis.

I think that if there's a reasonable chance we will need to reintroduce then we shouldn't RM. This probably is the case for utf8-string, but maybe not so for time. It's not *that* much effort to maintain the packages separately, but NEWing etc is a bit of a hassle if avoidable.

Iain

Attachment: signature.asc
Description: Digital signature


Reply to: