Why can that even happen ? What makes a difference between touch
creating a file and xpidl creating a file ? Why doesn't the latter have
a timestamp greater than the former ? *This* sounds fishy. Having to
wait there would still be a work around, not resolving the real problem
at hand.
The
Please, could you consider adding the sleep - it fixes the cause - rebuild
of some files during make install.
Weren't you saying the cause was the different variables values for make and
make install ?
Currently, my understanding of issue is as follows:
Files .done, which only signals
On Wed, Jul 09, 2008 at 08:07:44AM +0200, Petr Salinger wrote:
Please, could you consider adding the sleep - it fixes the cause - rebuild
of some files during make install.
Weren't you saying the cause was the different variables values for make and
make install ?
Currently, my
Hi,
I hopefully found the real cause of failure on GNU/kFreeBSD.
It is due to subsecond time resolution available in underlying
file system. The output of gcc does have subsecond timestamps cleared,
but output of touch preserves them.
This timestamp problem have been exposed by different
On Tue, Jul 08, 2008 at 06:51:16PM +0200, Petr Salinger wrote:
Hi,
I hopefully found the real cause of failure on GNU/kFreeBSD.
It is due to subsecond time resolution available in underlying
file system. The output of gcc does have subsecond timestamps cleared,
but output of touch preserves
This will probably never happen, because builds are likely to take
way more than than they currently do.
orion-bsd:/build/manual/iceape-1.1.10$ find . -name .done | wc
148 1486185
It happens 148 times per build, it adds 2,5 minutes,
the i386 buildd logs shows
Build started at
On Tue, Jul 08, 2008 at 09:31:14PM +0200, Petr Salinger wrote:
This will probably never happen, because builds are likely to take
way more than than they currently do.
orion-bsd:/build/manual/iceape-1.1.10$ find . -name .done | wc
148 1486185
It happens 148 times per build, it
7 matches
Mail list logo
