Hank Williams over at Why does everything suck? wonders how to best report an obscure and weird browser bug.  He’s got the right idea: save a static copy of the HTML, JavaScript, and CSS, and then strip out everything unrelated until you have the minimal chunk of code required to duplicate the bug. You can speed this up by making an educated guess at which code is related, and doing a binary search from there. Include the minimal test case with the bug report, and the Firefox (or WebKit) team will love you for finding a test case for such a wacky, rare bug.

I’ve done this a few times for Mosuki, and each time, I discovered a workaround before I got to reporting the bug. Each time, the bug was either the result of a interaction between components I thought were unrelated, or I found some other method of achieving the same result that didn’t trigger the bug. (Of course, it was usually an Internet Explorer bug that I was fixing, so there was no way to report the bugs.)