|
RE: ghc-pkg and -l: msg#00014lang.haskell.glasgow.bugs
> I just found out that the semantics of ghc-pkg has changed > between 5.04 and the recent HEAD: > > ~/source/gtk2hs:$ ghc-pkg-5.04 -f > /home/as49/source/gtk2hs/localpackage.conf -l > defaultPackage > > ~/source/gtk2hs:$ ghc-pkg-5.05 -f > /home/as49/source/gtk2hs/localpackage.conf -l > /home/as49/source/gtk2hs/localpackage.conf: > defaultPackage > /home/as49/local/lib/ghc-5.05/package.conf: > rts, base, haskell98, network, haskell-src, readline, > OpenGL, GLUT, > unix, lang, concurrent, posix, util, data, text, net, hssource, > greencard This is a side effect of ghc-pkg being rather more "correct" now, in the sense that it treats all the packages files as a stack in which later ones override earlier ones. Previously it only considered a single package file, which caused a number of problems (eg. dependencies which crossed package-file boundaries weren't possible). > This might be an improvement for the user but it is highly annoying in > shell scripts. I used to grep for gtk2 which now matches the > file path. Well, there are lots of ways to grep for a gtk2 package: ghc-pkg -l | grep -v : | grep gtk is one. > Furthermore I'd like to know if a certain package is installed in the > given package file, so I don't really want to see what's globally > available. > > If this verbosity is desired, could we have an -e flag which test if a > certain package exists? How about a -L flag which gives a list of packages in the given package file? Cheers, Simon
|
|
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| Previous by Date: | [ ghc-Bugs-734716 ] ghc-5.05: handle closed by hselect, SourceForge.net |
|---|---|
| Next by Date: | [ ghc-Bugs-738883 ] PosixTTY's testLocalFlag has c_iflags instead of c_lflags, SourceForge.net |
| Previous by Thread: | ghc-pkg and -l, Axel Simon |
| Next by Thread: | Re: ghc-pkg and -l, Axel Simon |
| Indexes: | [Date] [Thread] [Top] [All Lists] |
| News | FAQ | advertise |