Last night I wrote a simple parser for the Torque's database schema (it is quite nice despite the studlyCaps - more examples would probably show how useful it is an anger tho - I would be quite happy to use it to describe schema ) and tonight it didn't take long to finish the parser and rig it into Autodia ( 2 small mods to Autodia.pm, new entry in manifest, write the pod, make test, etc). an example Dia file can be found at the Autodia website
Hopefully I can start hacking on Codia and tedia2sql soon - it would be nice to use tedia2sql modules to a) improve the Dia parser in Autodia and b) provide SQL output for Codia (as well as perl and xml and stuff). The plan is start hacking at tedia2sql by providing Torque output and then start improving Autodia Dia SQL output so that you could snarf a database schema from SQL or Torque or DBI and then make any changes in Dia and then output to SQL using tedia2sql. This should hopefully take us some way towards schemamania's goal.
I have also added a couple of minor patches to the PHP and Perl handlers to improve their parsing - the Java version of Autodia is proving problematic - the INLINE::Java dependancy causes some problems (hopefully fixable using a begin block) and Java 1.4 breaks critical reflection features that the Java 'parser' uses. Following the old release early, release often rule you can find a pre-release of Autodia version 1.7 here - it compiles and I haven't had time to do a lot of testing so its a) not an official release and b) possible that it may not work very well if at all and c) patches are always welcome.
Looking at the progression of the DEVELOP and CHANGES file it is interesting to see that most of the new code has been in the handlers and very little at all has been changed in the core Autodia::Diagram code since version 1.0 and 1.2 (a year to the day and very nearly 2 years ago respectively). Also version 0.4 had 2000 lines over 7 files, version 0.5 had 2500 lines over 11 files, version 1.0 had 3000 lines of code over 11 files and version 1.7 pre-release has 5520 lines over 18 files and supports 4 forms of output and 8 different input types.
Seeing your project grow like that is such a nice feeling - it seems odd to think that I first released it to the world April 2001 and I still find it interesting and better still other people find it interesting.