Hello,
hopefully I'm just missing something, but I find the current handling of
"provides" in packages a little bit weird.
Let's say I'm building a subversion package of some library, like
libraw1394. I don't want to mix SVN and regular packages, so I name it
libraw1394-svn, and put both conflicts=('libraw1394') and
provides=('libraw1394') into the PKGBUILD.
When installing the package, pacman notices that it conflicts with the
original libraw1394 and asks me whether I want to remove it. Of course I
do as the new one provides the original libraw1394's functionality. But
then, pacman terminates lamenting that some dependencies will be broken
because there are packages that require libraw1394...
If it weren't for the "provides" field, I wouldn't wonder. But isn't its
purpose exactly this -- to prevent broken dependencies when a package
that provides some other's functionality is installed in place of the
original one?
And a similar problem appears the other way around as well. When trying
to replace the libraw1394-svn with 'pure' libraw1394, pacman once again
laments that some packages depend on the libraw1394-svn...
Am I missing something or is this a bug?
Best,
:g
_______________________________________________
arch mailing list
[email protected]
http://www.archlinux.org/mailman/listinfo/arch
