RESOLVED FIXED 55156
PatternAlternative leaked in YarrPatternConstructor::atomParenthesesEnd() 
https://bugs.webkit.org/show_bug.cgi?id=55156
Summary PatternAlternative leaked in YarrPatternConstructor::atomParenthesesEnd()
Michael Saboff
Reported 2011-02-24 09:23:58 PST
The removal of empty disjunction alternatives from the alternatives vector doesn't delete the removed alternative thus leaking it.
Attachments
Patch to delete unneeded PatternAlternative (1.46 KB, patch)
2011-02-24 11:06 PST, Michael Saboff 		oliver: review+
DetailsFormatted DiffDiff
View All Add attachment proposed patch, testcase, etc.
Michael Saboff
Comment 1 2011-02-24 11:06:30 PST
Created attachment 83682 [details] Patch to delete unneeded PatternAlternative
Oliver Hunt
Comment 2 2011-02-24 11:19:23 PST
Comment on attachment 83682 [details] Patch to delete unneeded PatternAlternative r=me
Michael Saboff
Comment 3 2011-02-24 11:25:56 PST
Committed r79594: <http://trac.webkit.org/changeset/79594>
Darin Adler
Comment 4 2011-02-24 11:58:14 PST
Comment on attachment 83682 [details] Patch to delete unneeded PatternAlternative View in context: https://bugs.webkit.org/attachment.cgi?id=83682&action=review > Source/JavaScriptCore/yarr/YarrPattern.cpp:500 > + PatternAlternative* altToRemove = parenthesesDisjunction->m_alternatives[i]; > parenthesesDisjunction->m_alternatives.remove(i); > + delete altToRemove; A couple post-patch-lading thoughts: I would prefer to do this with OwnPtr instead of an explicit delete. Maybe m_alternatives could actually use OwnPtr. I would also have called it alternativeToRemove instead of altToRemove.
WebKit Review Bot
Comment 5 2011-02-24 15:59:25 PST
http://trac.webkit.org/changeset/79594 might have broken GTK Linux 32-bit Release
Note You need to log in before you can comment on or make changes to this bug.