Blake generously offered to dive between this bullet and the president, in slow-mo yelling "guuuuuuhnnnnnnnnnnnn!", or was that "eeeeeeeeeeeeeefoooooooorrrrrrrrrrrrrxxxxxxxxxxxxxxxxxxxxxxxx!"? /be
*** Bug 328254 has been marked as a duplicate of this bug. ***
Our cycle detection wasn't handling lists, causing us to miss the "easy" case described here. Lists of length 1 are special in that you can do stuff on them that acts like you're doing it to its only element, therefore CheckCycle needs to deal. Note that we catch the case where kids->length > 1 before we ever get near the CheckCycle code.
Created attachment 212961 [details] [diff] [review] Fix
Comment on attachment 212961 [details] [diff] [review] Fix I was a collaborator on this fix, but sure, r=me. BTW, E4X sucks. /be
Fix checked into trunk.
Comment on attachment 212961 [details] [diff] [review] Fix approved for 1.8.9 branch, a=dveditz for drivers
> 1.8.9 branch 1.8.0, I mean
Fix checked into the 1.8 branches.
Checking in regress-328249.js; /cvsroot/mozilla/js/tests/e4x/Regress/regress-328249.js,v <-- regress-328249.js initial revision: 1.1 done I don't see a crash in today's ff trunk on winxp, but a current debug shell does appear to crash. I'll know more when I do a full test run.
Marking [rft-dl] (ready for testing in Firefox 126.96.36.199 release candidates) since in-testsuite+ indicates a test case exists in the js test library.
v ff 188.8.131.52/1.8/1.9 20060302 win/linux/mac