125562 - Regexp performance improvement

Status: VERIFIED FIXED

(Rhino Graveyard :: Core, defect)

Description

[Phil Schwartau]

This is the Rhino version of bug 85721 against SpiderMonkey. The testcase
for this bug is

              js/tests/ecma_3/RegExp/regress-85721.js

I have added this to the rhino-n.tests skip list for now, since it
contains timing tests that currently take a long time to complete.
Once the fix for this bug is made, they will run in a few milliseconds.

Comment 1

[Phil Schwartau]

Gee, I'm wondering if we have an issue here after all. On the above test,
my current time in either Rhino compiled or interpreted mode is 31ms :

java -jar D:/JS_trunk/mozilla/js/rhino/build/rhino1_5R4pre/js.jar
-f ecma_3/shell.js
-f ecma_3/RegExp/shell.js
-f ecma_3/RegExp/regress-85721.js


Note: I'm using WinNT; 128M RAM, 500MHz CPU.
Norris, do you get about the same time on this test? 

Comment 2

[rogerl (gone)]

(embarrased cough) I think what you're seeing here is that there are still bug 
fixes from SpiderMonkey that have yet to be transferred to the Rhino engine and 
those fixes will promptly whack the performance down. The engine rewrite in the 
attachment to #85721 fixes the bugs and gets the performance back in line. 

Comment 3

[rogerl (gone)]

Fix checked in - new engine implementation ported from SpiderMonkey.

Comment 4

[Phil Schwartau]

Verified Fixed - the above testcase now passes in the rhino, rhinoi shells.

Comment 5

[Igor Bukanov]

Targeting as resolved against 1.5R4