Back in ComboBox hell, and finally read this: https://bugs.adobe.com/jira/browse/SDK-25705
I've always had problems with comboboxes and binding, but didn't realize this particular bug was *introduced* in 3.5, rather than having been there all along. Of course I updated our application to use 3.5 because of bad bugs in charting. I'm glad bug fixes continue in the 3.x branch, but it's absolutely critical that there are no regressions. So while I hope 3.6 is imminent, let's hope it's well-tested. I trust the testing infrastructure and architecture has been improved for 4.x.
