While trying to track down the root cause of bug 516334, noticed a typo in nsJPEGDecoder.cpp: the code uses && where & is intended (testing a bit flag). This could potentially cause it to fail to issue appropriate notifications, although fixing this does not actually resolve 516334 (which is not JPEG-specific but can occur with any image format).

Right now this is harmless, because there's only 1 possible flag, so & and && are equivalent - so the net effect is if (mFlags), which will be either 1 or 0 depending on whether the one flag is set. http://mxr.mozilla.org/mozilla-central/source/modules/libpr0n/public/imgIDecoder.idl#68 Definitely needs to be fixed though.

Yeah, that's consistent with the observation that fixing the test didn't make any difference to the problem in 516334; it just happened to be a line of code that got looked at along the way. (I hadn't checked whether there were other possible flag values; but did notice that in all the traces I looked at, mFlags was zero anyway.)

Comment on attachment 414819 [details] [diff] [review] fix the flag test to use bitwise AND rather than logical AND r+

Pushed to trunk: http://hg.mozilla.org/mozilla-central/rev/f5ae3ffa3363