[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Bug#324030: swt-gtk: FTBFS on 64 bit arches: error: cast from 'void*' to 'jint' loses precision

On Tue, Sep 13, 2005 at 10:52:02AM -0600, Shaun Jackman wrote:
> The build times out on m68k:
> make[1]: Entering directory `/build/buildd/swt-gtk-3.1'
> gcj -fPIC -fjni -c `find org -name *.java`
> make[1]: *** [java-stamp] Terminated
> make: *** [build-stamp] Terminated
> Build killed with signal 15 after 150 minutes of inactivity
> What's the solution?

That was me. The build was running on one of our fastest machines, but it
did not produce any output for two and a half hours. This is often a sign
for a crashed build. Since we do not want our buildds to be blocked by this,
we have a default timeout of 150 minutes. If there has been no output to the
build log file, the build is killed. Of course we can increase the timeout,
but it would be good to know some reasonable numbers, especially since the
build could end up on a slower machine and still timeout. This would only be
a fix, a solution would be if there is some output in reasonable time
intervals so we know that the build is still doing something. How about
compiling the source files separately, there would a new log line for every
source file? Some package like gcc are producing output at regular intervals
just to prevent the build from being killed:

: # start logwatch script for regular output during test runs
chmod +x debian/logwatch.sh
debian/logwatch.sh -t 1800 -p
/build/buildd/gcc-3.2-3.2.3ds9/build/logwatch.pid \
        -m '\ntestsuite still running ...\n' \

But I still think a real solution would be some visible sign of build
progress for the sake of the nervous buildd maintainer.


Reply to: