> Glancing at the code, it looks like one can simply keep a pointer to the > list and add elements to it, but, again, I've not yet tested this. If you > need to rely on this feature, we should add a unit test, to make sure it > continues to work.
Relying would be a bit strongly put for my case, but I would find it elegant. :-) I'll create a test for this once the current test attached to AVRO-95 (which tests that writing multiple types through reflection to one file works at all) shows a green bar. Cheers, Eelco
