I feel sorry for Eric, but I had absolutely no choice; his code was an absolute disaster and would have caused millions of dollars of lost developer time as they dealt with the crashes it caused, losing a major feature of the debugger in the process. It was a clear call, in my opinion.
So with two days before the deadline, we decided to cut the feature; he'll try again next milestone, with enough time to write actual unit tests, redo his thread model, and solve the real problems.
It's amazing how much more relaxed Eric is now. With no more panic to hit a deadline, and no more fear of failure because it already happened, he's in much better shape. It's totally understandable, and in my opinion healthy for him- what's done is done, and he's immediately moving on.
It helps me not feel bad about having to cut the feature, but, well, I'm not the one who wrote all those bugs.
I've migrated to DreamWidth. The original post is at http://kistaro.dreamwidth.org/456843.html. View comments at http://kistaro.dreamwidth.org/456843.html#comments ; go ahead and use OpenID to post your own.