334795 - RegEx exec() maintains state even though regex gets reinitialized

Status: RESOLVED DUPLICATE of bug 98409

(Core :: JavaScript Engine, defect)

Description

[Christian Wenz]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; de; rv:1.8.0.2) Gecko/20060308 Firefox/1.5.0.2
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; de; rv:1.8.0.2) Gecko/20060308 Firefox/1.5.0.2

Evaluating regular expressions with exec() and using the "g" modifier maintains state although the regular expression get initialized again. Even worse, the number of matches from the first exec() call is maintained for subsequent exec() calls.

Reproducible: Always
Steps to Reproduce:
1. Load Attachment (it creates the regex /(a*(b*))c+/g and than calls exec() with parameter "abc c"; then, the returned array is shown on the screen)
2. Click on the button. 
3. Click on the button again.
4. Note the results

Actual Results:  
After the first click: 0: abc  1: ab  2: b
After the second click: 0: c  1:  2:  

Expected Results:  
0: abc  1: ab  2: b after both clicks
Note: even when the second click should return the second match (which in my opinion is not the desired behavior, since the regular expression gets instantiated again), the length of the array returned by exec() should be 1 and not 3. The value 3 is taken from the first result (changing the regex proves that).

Comment 1

[Christian Wenz]

Attachment: testcase to reproduce the issue

Comment 2

[:Gavin Sharp [email: [e-mail]]]

see also bug 237111

*** This bug has been marked as a duplicate of 98409 ***

Comment 3

[Christian Wenz]

(In reply to comment #2)
> see also bug 237111
> *** This bug has been marked as a duplicate of 98409 ***

aah, didn't find this. Thanks! So the fact that the literal regex maintains state is a dup and will be fixed once ECMA-262 edition 4 is implemented. But what about the array of matches keeping the same size? I think this is not covered in the bugs you mentioned. What do you think?

Comment 4

[Andreas]

The length of the array depends only on the number of capturing parentheses,
no matter if they matched or not (here they even matched the empty string):

js> var re1= /(a*(b*))c+/g
js> re1.exec( "abc c").toSource()
["abc", "ab", "b"]
js> re1.exec( "abc c").toSource()
["c", "", ""]
js> var re2= /(a(b)*)*c+/g
js> re2.exec( "abc c").toSource()
["abc", "ab", "b"]
js> re2.exec( "abc c").toSource()
["c", undefined, undefined]