Bug#480846: Err, this is not a make issue.
On Tue, May 13, 2008 at 04:49:11AM -0500, Manoj Srivastava wrote:
> reassign 480846 glibc
> thanks
>
> Hi,
>
> >> We skip the test suite on hurd-i386, and it appears the inferior time
> >> resolution on hurd-i386 results in some time skew with respect to the
> >> build_libc and check_libc targets, which in turn makes it redo the whole
> >> build (why that is is another question, but at least AFAICT [build log
> >> not yet available on buildd.debian-ports.org] it really starts over) for
> >> the install_libc target in binary-arch:
>
> This is a race condition, and needs to be fixed, perhaps with a
> sleep 1 in appropriate places.
>
> > This is a problem with make. Switching back to version 3.81-3 fixes
> > the problem. Reassigning the bug.
>
> It really is not. What you have is an intermittent problem,
> which will trigger only sometimes. There was no change between 3.81-3
It is not an intermittent problem. It is something 100% reproducible
on at least gcc and glibc. When switching back to version 3.81-3, the
problem fully disappears. I haven't tried yet the version 3.81-3.1, nor a
rebuild of version 3.81-4 in the current environment.
> and 3.81-4 which would make a difference in resolving the time.
There are 1.5 years between the too version. That is a huge difference,
the whole build environment has changed.
--
.''`. Aurelien Jarno | GPG: 1024D/F1BCDB73
: :' : Debian developer | Electrical Engineer
`. `' aurel32@debian.org | aurelien@aurel32.net
`- people.debian.org/~aurel32 | www.aurel32.net
Reply to: