RT $VERSION got botched to 0.0.0 in a homegrown rpm


I rolled an rpm version of 5.0.1 for internal use on fedora. and it looked fine.

After creating a bunch of queues and tickets while practicing using the fine tool, noticed that somehow the version was showing up in all the screens as 0.0.0.

Poked around, and discovered that indeed, $VERSION = ‘0.0.0’ in /usr/share/perl5/vendor_perl/RT/Generated.pm.

Thought, ok, that should be a simple fix, just set the correct version and all will be good. BZZZT. nope.

Restarting apache succeed, but the refresh of my “RT at a glance” home page threw a 500 error. So reverted my change, and now it’s ok again.

Is there something intrinsic about the version number that gets put in the DB or somewhere that I need to fixup?

There are only about 100 tickets, that wrote to myself basically, but they were items related to bootstrapping a business, so it’s not THAT big of deal to completely start over with the fixed $VERSION, but where could I look to tweak things to repair my self inflicted wound?

I suppose I have 2 specific questions:

  1. Is there a way to correct the database if that is the issue?
  2. Other than a single reference to lib/RT/Generated.pm in configure.ac, how does the correct version end up in Generated.pm?


EDIT: Oh, the sed during configure did not set ANY @RT_VERSION@ and others to anything but 0. Everything has 0.0.0. gah.

EDIT2: simplez. just edit the Makefile.in to use a restyled version of the SNAPSHOT to create the snapshot, and rework the spec file to use that info.

The answer to my questions, so far, is that having the $RT:VERSION correct in all the files appears to have everything working, no database touches needed. Although I did back up the DB prior to updating the new rpm.

If anyone is interested, can supply the deltas, but since this is a one off case for my environment, will hold off posting on the details.

Now, I have to go and change the bandages on my bloody foot.