On 01/06/11 00:47, Dan McGee wrote:
On Mon, May 30, 2011 at 4:11 PM, Andreas Radke<a.radke@arcor.de> wrote:
Am Tue, 24 May 2011 16:26:11 -0500 schrieb Dan McGee<dpmcgee@gmail.com>:
On Tue, May 24, 2011 at 4:01 PM, Andreas Radke<a.radke@arcor.de> wrote:
Am Sat, 5 Feb 2011 17:10:16 -0600 schrieb Dan McGee<dpmcgee@gmail.com>:
It rests in [testing] because it is totally broken with handling MAKEFLAGS, and upstream is totally unresponsive. Allan isn't looking for signoffs... https://savannah.gnu.org/support/index.php?107487
Things are getting worse. New LibreOffice 3.4 will require make 3.82 to compile. I can do this in testing chroots when there's no other .so related stuff though.
So far I've only seen the kernel compiling non-multithreaded. Did you read the bug report, by chance? I was able to make it 100% reproducible compiling git. https://savannah.gnu.org/support/index.php?107487#comment0
Fedora keeps applying that patch in 3.82 - could this solve your test case? Gentoo has dropped it seems.
http://pkgs.fedoraproject.org/gitweb/?p=make.git;a=blob;f=make-3.82-jobserve...
I don't think so, unfortunately. That patch has been around since 3.81 which is known-working for this particular bug, and the code it is dealing with looks like not the stuff Allan and I were digging around in for a bit.
The jobserver stuff is different. Note that Fedora (at least v14) is affected by this bug too. Anyway, I am happy for this to be moved to [core] at any stage. I obviously never considered it broken enough to remove from [testing]... If no-one objects in the coming days, it will be moved. Allan