/var/tmp/portage/dev-lang/ruby-1.8.6-r1/image/usr/lib/libruby18.so.1.8.6 [ 1] .hash HASH 00000114 000114 0017dc 04 A 3 0 4
Maybe you could elaborate on how did you come to the conclusion that your LDFLAGS are ignored, which ones are ignored and where's the problem. The above cryptic stuff unfortunately doesn't explain anything.
(In reply to comment #1) > Maybe you could elaborate on how did you come to the conclusion that your > LDFLAGS are ignored Some LDFLAGS can change existence of some sections in ELF files. I have such (at least) 2 flags in my LDFLAGS and one of them always change existence of some sections when isn't ignored. I changed /usr/lib/portage/bin/prepstrip, so that it now additionally runs readelf and grep and copy the output to the log file. I perform some tests (reading emerge logs) and >99,5% of results are correct. > which ones are ignored All. > and where's the problem. Ignoring LDLFAGS increases files sizes and decreases performance.
Created attachment 121677 [details, diff] Patch
As of ruby-1.8.6, we can set EXTLDFLAGS=$LDFLAGS to feed user LDFLAGS to ruby. Something like this: --- ruby-1.8.6-r1.ebuild 2 Jul 2007 14:50:46 -0000 1.14 +++ ruby-1.8.6-r1.ebuild 23 Aug 2007 14:19:43 -0000 @@ -86,7 +86,7 @@ --with-sitedir=/usr/$(get_libdir)/ruby/site_ruby \ || die "econf failed" - emake || die "emake failed" + emake EXTLDFLAGS="${LDFLAGS}" || die "emake failed" } src_test() { Arfrever, can you check if the trick works for you?
(In reply to comment #4) > Arfrever, can you check if the trick works for you? It works.
ruby-1.8.6_p36-r4 has been added to the tree. Hopefully it won't pull a new stream of bugs due to bad LDFLAGS usage :)