On 27.03.2020 18:42, Martin Husemann wrote: > On Fri, Mar 27, 2020 at 06:36:29PM +0100, Kamil Rytarowski wrote: >> I tested the code without our local patch as mentioned in my commit, >> even in the snapshot 1 commit before jak@'s change. >> >> In all the cases I get P_tmpdir defined and pointing to "/tmp/". >> >> I have not tested it on NetBSD-8 as host. All my tests were with NetBSD >> 9.99.51 as host with GCC style distribution. > > We do not care that much for -8 (even not enough to have that patch you are > backing out pulled up). I only used it to test as that was what I had at hand > w/o the "force /tmp" commit. >
If this words out of the box with P_tmpnam on newer NetBSD it might be a good enough reason to backout the patch. > If libiberty has been fixed upstream and the fix is not needed any more: > great! > I will try to research why I get P_tmpnam defined. > From your description it was not clear to me if you did a native gcc build > (./configure && make or something) or used our in-tree reachover > infrastructure > (which might be what was causing the bug). > I tested all GCC/GDB build modes: - in pkgsrc - standalone (./configure && make) - ./build.sh tools - ./build.sh generated distribution In all of them P_tmpdir worked for me. > Martin >
