The idiom for (LInsHashKind kind = LInsFirst; kind <= LInsLast; kind = nextKind(kind)) will trigger comparison is always true due to limited range of data type warnings/errors dependings on compiler configuration.
Created attachment 418416 [details] [diff] [review] Patch
Comment on attachment 418416 [details] [diff] [review] Patch "compilers that will insist..." is heresay and spreads fear, uncertianty, and doubt... the comment should say what compiler it is, so later we can re-check whether the hack is still needed.
I wrote "will trigger... warnings/errors depending on compiler configuration", "compilers will insist"... :-) For the record, it was a variant of gcc 4.4 building for Android; I'm guessing (without proof) that any gcc 4.4 variant can be persuaded to complain in similar fashion if you configure its warning level in a similarly anal-retentive manner.
Comment on attachment 418416 [details] [diff] [review] Patch nj-c: http://hg.mozilla.org/projects/nanojit-central/rev/e2a738ba767e
(Fixed 18 months ago) http://hg.mozilla.org/mozilla-central/rev/fbf83c1515fa