Twice in the past couple of weeks we've had JSTRACE things added, for JS scripts and type inference, without cases being added to JS_PrintTraceThingInfo, which causes JS_DumpHeap to assert. We should add some kind of test to check for that, either some kind of JS_assert, or a crash test that calls JS_DumpHeap. I guess that since this is just debug logging code, so we could alternatively just fail more quietly. But that sounds bad.
As my patch for the bug 674251 has the same problems with JSTRACE_SCRIPT I am going to update the patch there and add JSGCTraceKind enum. Then at least with GCC we get warnings about missing labels.
Yeah, that sounds like a much better fix than anything I had thought of!