diff options
| author |   Mart Raudsepp <leio@gentoo.org> | 2008-09-27 09:16:32 +0000 |
|---|---|---|
| committer | Mart Raudsepp <leio@gentoo.org> | 2008-09-27 09:16:32 +0000 |
| commit | 0f4bb94e0ac1f20ab811132f479b68183b010306 (patch) | |
| tree | d73f3b437fdea41bc29b1bb42f2aa0e6ea521fc7 /dev-util/bzr | |
| parent | Automated update of use.local.desc (diff) | |
| download | gentoo-2-0f4bb94e0ac1f20ab811132f479b68183b010306.tar.gz gentoo-2-0f4bb94e0ac1f20ab811132f479b68183b010306.tar.bz2 gentoo-2-0f4bb94e0ac1f20ab811132f479b68183b010306.zip | |
Sort the list of Makefile.am and Makefile.in files used in gnome2_omf_fix(), so that
it is ensured the .am one is listed before the respective .in for maintainer mode
regeneration to not kick in due to modtime of .am being newer than .in and the listed files
being possibly patched with sed for a scrollkeeper fix.
This maintainer mode build file regeneration is reliably reproducable only if PORTAGE_TMPDIR
is on a filesystem with jiffies-granularity timestamps - tmpfs in particular. In the common
case (ext3 and reiserfs) it is only second-granularity and GNU make rule dependency for
am--refresh doesn't trigger almost never because the timestamp is equal unless it hits a
second boundary pretty exactly in the middle. This should fix bug 236649 and all the few dozen
INVALID bugs Diego Pettenò has filed against GNOME packages that claim the maintainer mode
rebuild is caused by patches in $FILESDIR, while it must have been simply the omf_fix combined
with tmpfs usage.
Diffstat (limited to 'dev-util/bzr')
0 files changed, 0 insertions, 0 deletions