RESOLVED FIXED 15998
Abstract much of RMATCH into MatchStack functions
https://bugs.webkit.org/show_bug.cgi?id=15998
Summary Abstract much of RMATCH into MatchStack functions
Eric Seidel (no email)
Reported 2007-11-15 03:25:03 PST
Abstract much of RMATCH into MatchStack functions I'm preparing to further cleanup "recursion" in match(), starting by making RMATCH() much much smaller. Currently I'm inlining the frame management (since that's how it was before). I'm not sure that will be a win in the end. This patch actually caused my SunSpider results to stabilize, showing no slowdowns, a 9.3% improvement in regex-dna, and a 1.0% win over all (combined improvements from this and bug 15996)
Attachments
abstract RMATCH into MatchStack functions (1.78 KB, patch)
2007-11-15 03:25 PST, Eric Seidel (no email)
no flags
abstract RMATCH into MatchStack functions (8.76 KB, patch)
2007-11-15 03:26 PST, Eric Seidel (no email)
sam: review+
Eric Seidel (no email)
Comment 1 2007-11-15 03:25:58 PST
Created attachment 17290 [details] abstract RMATCH into MatchStack functions
Eric Seidel (no email)
Comment 2 2007-11-15 03:26:33 PST
Created attachment 17291 [details] abstract RMATCH into MatchStack functions
Sam Weinig
Comment 3 2007-11-15 10:51:36 PST
Comment on attachment 17291 [details] abstract RMATCH into MatchStack functions The typedef should not be necessary in c++. +typedef struct matchframe { nor should this explicit struct + struct matchframe* prevframe; r=me.
Eric Seidel (no email)
Comment 4 2007-11-15 16:15:02 PST
Landed.
Note You need to log in before you can comment on or make changes to this bug.