(I think I really do understand the different ways of manipulating the BTS now - this message is virtually identical to the one I sent to control@ a few minutes ago.)
I originally sent the wrong BTS command so the bug wasn't properly reopened. This bug no longer manifests the same way because the beagleindex user is only removed on package purge, however the daily cron script still runs and only seems to succeed silently because stderr is redirected to /dev/null when trying to execute the command (in tools/beagle-crawl-system.in:67). I think it is dangerous to rely on this behaviour and an explicit check should be made at the top of the script to see if the beagle-build-index command is available. The attached patch tests for the existence of the beagle-build-index command before continuing. I also think it's probably wise to call "set -e" at the top of the script so that if a command fails (for instance the chown on line 37), the script will not continue executing. I have not added this in the attached patch. Cheers, -Ted
--- tools/beagle-crawl-system.in 2006-06-20 03:54:27.000000000 +1000 +++ tools/beagle-crawl-system.in-new 2006-08-27 02:03:35.000000000 +1000 @@ -26,6 +26,10 @@ # SOFTWARE. # +if [ ! -x @sbindir@/beagle-build-index ]; then + exit 0 +fi + umask 022 CRAWL_USER=beagleindex
signature.asc
Description: OpenPGP digital signature
