Our development process is highly automated via a raft of bash and php scripts (including subversion hook scripts.) These scripts do a number of things to integrate with our Bugzilla 3.0 installation.
But the current integration approach is a bunch of SQL calls which update the bugzilla database directly - which obviously has a number of downsides - including making me nervous about upgrading to 3.2 in case the database schema has changed!
For example, to add a comment to a bug, I'm doing an INSERT into the longdescs table.
So my (slightly long-winded) question is:
According to the Bugzilla WebServices API, some of the features needed (e.g. changing a bug status) are not yet available, so for the time being, direct SQL calls seem the most appropriate option.
The database schema has not changed significantly between version 3.0 and 3.2 so this is a practical way forward.