Summary: | Support for libffi in toolchain.eclass | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | David Leverton <levertond> |
Component: | Eclasses | Assignee: | Gentoo Toolchain Maintainers <toolchain> |
Status: | RESOLVED FIXED | ||
Severity: | enhancement | CC: | bo.andresen, flameeyes, peper |
Priority: | High | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Bug Depends on: | |||
Bug Blocks: | 160183, 163724, 198875, 200679, 209673, 209675 | ||
Attachments: | patch |
Description
David Leverton
2007-11-21 00:12:10 UTC
Created attachment 136562 [details, diff]
patch
Tested with gcc 3.4.6-r2 and 4.1.2 on amd64.
The toolchain folks know the eclass far better than me, so they may want to change things a bit if there's anything done inappropriately, but the important parts are the sed and the change to the ffitarget.h move to make it happen with the new USE-flag.
This will cause the build of most gcc 3.4.x versions to fail if the objc USE-flag is set, because those ebuilds try to make the same change again. I'm not sure what the best solution for that is, so I didn't try to do anything in the patch, but if the objc-implies-libffi logic is considered desirable, it probably should be moved into the eclass as well.
*** Bug 209675 has been marked as a duplicate of this bug. *** ive added it ... the libtool cruft will have to wait for a general solution # cat toolchain.eclass [...] is_libffi() { has libffi ${USE} || return 1 use libffi } [...] s/USE/IUSE/. thanks, ive fixed that now http://sources.gentoo.org/eclass/toolchain.eclass?r1=1.361&r2=1.362 |