I've only looked at the glibc-2.5 ebuild, but the others may be broken in the same way. In check_nptl_support(), want__thread is called like so: if ! eend $(want__thread); then This doesn't seem to work, as it passes the check on hppa with gcc-4.1.1-r3, when the check really fails. If I change it to the following, then it works properly: want__thread if ! eend $?; then
Also, I only found this because I removed the explicit 'return 1' in want_tls() for hppa to attempt to build glibc-2.5 with nptl.
the point was to merge the steps, not break them apart again i think we actually want: if ! eend $(want__thread; echo $?); then
Sure, that would work as well. It doesn't really matter as long as the code actually does what it was originally intended to do ;)
Not that there was any doubt, but that variation fails properly.
fixed in cvs, cheers