Re: [gentoo-alt] readline ebuild problem

[Alan Hourihane]

On Wed, 2008-11-26 at 16:37 +0000, Alan Hourihane wrote:
> On Wed, 2008-11-26 at 17:26 +0100, Michael Haubenwallner wrote:
> > On Tue, 2008-11-25 at 17:09 +0000, Alan Hourihane wrote:
> > > On Tue, 2008-11-25 at 18:05 +0100, Fabian Groffen wrote:
> > > > On 25-11-2008 17:02:11 +0000, Alan Hourihane wrote:
> > > > > On Tue, 2008-11-25 at 17:57 +0100, Fabian Groffen wrote:
> > > > > > On 25-11-2008 16:53:19 +0000, Alan Hourihane wrote:
> > > > > > > There's a bug in the readline ebuild that says this...
> > > > > > >
> > > > > > > [[ $(get_libname) != .a ]] &&
> > > > > > >
> > > > > > > It should be....
> > > > > > >
> > > > > > > [[ $(get_libname) == .a ]] &&
> > > > > >
> > > > > > I think haubi did this on purpose, as AIX has .a for "shared libraries"
> > > > > > :(
> > > > > > It moves the "static" library out of the way, but if the platform
> > > > > > happens to use .a for that, it refrains from doing so.
> > > > > >
> > > > > > Haubi has the last word on this.
> > > > >
> > > > > Then I need to add....
> > > > >
> > > > > [[ ${CHOST} == *-mint* ]] &&
> > > > > mv "${ED}"/$(get_libdir)/lib{readline,history}.a
> > > > > "${ED}"/usr/$(get_libdir)/
> > > >
> > > > Hahaha... I knew this was going to break some time...  Let's see what
> > > > haubi thinks/wants.  Does mint have shared libraries, or just static
> > > > ones?
> > >
> > > Just static.
> >
> > Ohw, no shared libs!
> > Wouldn't this require some (portage-)mechanism to perform a
> > reverse-dependency-rebuild based on {,R}DEPEND?
> >
> > As far as I understand, get_libname() returns the extension used for
> > *shared* libraries (which on AIX unfortunately can be '.a').
> >
> > What should $(get_libname) return when there are no shared libraries?
> >
> > The test in readline just prevents moving 'lib*.a' out of /lib when it
> > (potentially) is the shared library.
> >
> > When there are no shared libs at all, isn't it just wrong to have
> > get_libname() returning '.a'? Maybe it should return either the empty
> > string, or something like '.irrelevant' IMO.
>
> Having get_libname() return .a on FreeMiNT makes things behave correctly
> for the majority of ebuilds.
>
> Given AIX uses .a for shared libs makes me think that readline should
> really be AIX specific and use CHOST == *-aix* rather than just globally
> using != .a.

Any decision ??

Alan.