Summary: | gcc-3.1-r7 / libtool-1.4.1-r8 / kde libtool problems... | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Daniel Robbins (RETIRED) <drobbins> |
Component: | [OLD] Development | Assignee: | Brandon Low (RETIRED) <lostlogic> |
Status: | RESOLVED FIXED | ||
Severity: | blocker | CC: | azarah, terje |
Priority: | High | ||
Version: | unspecified | ||
Hardware: | x86 | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- |
Description
Daniel Robbins (RETIRED)
![]() Hmms, I was scared that changing $libdir will have some impact. I guess we can fix it one of three ways: 1) leave the libdir sed as it is now, but install the .la files in /usr/lib/ AND /usr/lib/gcc-lib/i686-pc-linux-gnu/3.1/. 2) Move the .la's to /usr/lib/gcc-lib/i686-pc-linux-gnu/3.1/ and leave the libdir sed as it is now 3) Dont modify $libdir, but still move the .la's to /usr/lib/gcc-lib/i686-pc-linux-gnu/3.1/ as with the dual installation which seem to work fine with all problems that was experienced with the single install. Anyhow, whatever should be tested well I guess before it is changed again. I am certainly not an expert on these libtool issues, but would it be possible to have libtool always check /usr/lib as a fallback if the .la file doesn't exist in the specified location? I also think it would be very wise to see how other distributions like Mandrake and Red Hat handle these issues. libtool: link: warning: `/usr/lib/libstdc++.la' seems to be moved libtool: link: warning: `/usr/lib/libstdc++.la' seems to be moved When I copied the .la files to the /usr/lib/gcc-lib/i686-pc-linux-gnu/3.1/, I got those warnings but everything seemed to work. It may indicate that the solution I chose is not optimal (we don't want warnings like that) the -r7 ebuild currently in rsync does what you did to fix it (as I mentioned in #gentoo-dev) I'm going research as to how other distros handle this issue, and if I find a better fix I'll -r8 it. I'm leaving this bug open so that users will see it just in case they got the bad -r7, and to remind me to find a better fix. Well, -r6 is pretty much what Mandrake does, and I guess then Redhat as well. This setup allows multiple installations, and did work well with 3.0.4 and gcc-3.1 except for gimp-print. I think that maybe we should check how they handle gimp-print, as it could be something that is broken in there, rather than our gcc-3.1-r6 ebuild. *** Bug 4493 has been marked as a duplicate of this bug. *** looks like azarah has dealt with this in gcc-3.1-r8 which is masked for testing still. |