It was a bit more involved than just a one liner, especially the functional test part.
Also, in real life conditions, whether or not this feature will actually work now depends on a large part on external factors: which browser is used, which authentication method (and how the authentication data is saved), which web server, etc.
For example in my tests with tracd and digest auth, I managed to make it work by saving the username in utf-8 and computing the hash on that, but only with Chrome and Opera, not with Safari, FF or IE10. IE10 sends back the username in latin1, so no wonder, but FF sends it back in utf-8 so the error must be elsewhere in that case (well, it's known not to work anyway). And even for Chrome, though it "workedforme", there's still an open issue about this, so YMMV.