Feature #2428: upgrade to ruby 1.9.3
Okay so I dug in and I've finally pursuaded all of the tests to pass. I did not go out of my way to create new tests, but I did fix every existing test and deprecation that was possible to fix, and in many controller classes the test coverage was significantly increased (from default generated and completely failing tests to at least a reasonably complete set of RESTful tests). It was a pretty big job, but it probably fixed at least a couple security holes and bugs in Tapestry. Mostly I just wanted to finally get the tests into a condition where we'd notice if something went awry, going forward. It's hard to tell the difference between 376 tests failing and 379 tests failing, but the difference between 0 and 3 is huge!
I did notice thought that I probably spent too much time fixing controllers and their tests that may not even still be in use. There are so many dark corners to Tapestry!
In any case, now we can choose to increase test coverage - the safer but more tedious option - or move on to upgrading to ruby 1.9.3 and 2 and also upgrading Rails upwards.
I have not looked into tools for analyzing test coverage of code - perhaps this might be a good way to measure it. But certainly it would be good to know whether a lot of the more obscure controllers that don't seem to have active enrollment steps associated with them should still be maintained - or removed.
(There are still some deprecation notices about some certain shoulda directives but it's just because we're caught between an old version of Ruby and newer gems, of course.)