Ignore the 2nd patch set; sorry about that. So, Thomas, I took your suggestion from the issue tracker and am assuming the 1st value of acceptable values--but only if the user has not already set an explicit value.
I think this is an acceptable compromise. It keeps the existing behavior of "setValue(null)" adding an entry if needed (added a test for that). Ironically, 3 of the 4 existing ValueListBoxTest test methods were failing for the same mistake from the bug report--the tests didn't think "null" was going to show up as an extra value. If our own unit test got confused, I think that is a good hint that the "call setValue before calling setAcceptableValues" convention of the current implementation is less than ideal. I should add an assertion that the 1st value gets picked, and probably some javadocs...but other than that, what do you think, Thomas? http://gwt-code-reviews.appspot.com/1619803/ -- http://groups.google.com/group/Google-Web-Toolkit-Contributors
