Doing a clean/full run on lintian.d.o

[Niels Thykier <niels@thykier.net>]

Hey,

As the subject says, I started a full run today.  Originally I had
intended to just upgrade, do an incremental run to see it worked and
logout... However, I managed to turn it into a complete disaster.

I "tailed" the lintian.log and realized it was spewing out errors.  The
problem was clean_env (in lib/Util.pm) dying due to missing locale and I
eventually fixed it by generating an en_US locale in
$LINTIAN_ROOT/locale.  I guess I managed to break said locale myself.

By then I had managed to corrupt part the lab info files
(binary-packages was empty) and html_reports had been aborted half-way
through leaving lintian.d.o without index.html.

I fixed the latter by feeding html_reports with the old lintian.log.
While it did restore the index site and all, it had a "minor" issue
looking up "a couple of maintainers"[1].
  By now the package import is complete and I reran html_reports (with
same log).  It gave fewer errors/warnings this time, but you may still
see some "hick-ups".


Lintian is currently running in screen (as the lintian user)[2] and have
past src:aspell not long ago.  Assuming the clean run is similar to a
full run, lintian should be done with the archive check in 3 days time.

~Niels

[1] By "minor" I mean "major" and by "a couple of maintainers" I mean
"most of them".

[2] Accessible via:
$ sudo -u lintian -s
$ script /dev/null
$ screen -x

harness runs in window 0.  harness.log and lintian.log are tailed in
window 1 and 2.