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

Re: Bug#504528: libghc6-configfile-dev: Fails to configure: MissingH-1.0.1 doesn't exist



Joachim Breitner wrote:
>> Oh drat, you're right.  I goofed on that.  OK, so this would have been
>> fine if I hadn't broken my own scheme then, eh?  If cabal said just
>> 1.0.2, it would just work?
> 
> No, it’s just fine:
> libghc6-missingh-dev (>= 1.0.2.1), libghc6-missingh-dev (<< 1.0.2.1+)
> means essentially (== 1.0.2.1), as long as you don’t start to add a 5th
> digit. So when you upload a new version (with a new cabal version), the
> old packages become uninstallable until re-built, just as intended.

I think you're missing the point.  I *want* it to accept 1.0.2.*,
because revisions at that level only signify Debian changes.  Or would
have, had I not goofed in .cabal.

> I’m currently talking to kaol on #debian-devel, and agrees, if I got him
> correctly.
> 
> BTW, John, while I’m at it I found out that there are more haskell
> packages that have unversioned binary dependencies, some of them are
> yours:

[ snip ]
> 
> And furthermore the configfile-doc package does not have it’s
> haddock-file registered with ghc-pkg and is not liked from
> file:///usr/share/doc/ghc6-doc/libraries/index.html (comparing it to
> haskell-mtl). Is that just not-yet-done, or is there a specific reason
> for these variations?

You're quite right on all of that.  Generally speaking, I have
unfortunately not had a lot of time to follow the current progress on
the documentation front in the last year or so.  It's not just
ConfigFile; it's all my packages.  This is all part of my (way too long)
to-do list.  When I first packaged these things for Debian, that
mechanism didn't yet exist.

-- John


Reply to: