504567 - String.split() fails when using patterns that can match empty string and pattern is not at the beginning of string

Status: RESOLVED WONTFIX

(Tamarin Graveyard :: Virtual Machine, defect)

Description

[Brent Baker]

Transferred bug from Adobe JIRA bug system: 
https://bugs.adobe.com/jira/browse/ASC-3628

String.split(pattern) will only match if the pattern is at the beginning of the string.

Test:
trace("ab".split(/a*/));
trace("abab".split(/a*/));
trace("babab".split(/a*/));

Actual Results:
,b
,bab
babab

Expected Results:
,b
,b,b
b,b,b

Expected results are produced by firefox

Comment 1

[Brent Baker]

Attachment: testcase

Comment 2

[Brent Baker]

Attachment: testconfig to expect failure of the split() tests

Comment 3

[Brent Baker]

This is not a recent injection, has been around since fp9

Comment 4

[Brent Baker]

testcase pushed with "expectedfail" 2154:1103df25c6a4

Comment 5

[Michael Daumling]

I don't believe that PCRE can be configured to work as expected. ECMA-262 15.5.4.14 says:

"If separator is a regular expression, only the first match at a given position of the this string is considered, even if backtracking could yield a non-empty-substring match at that position."

In the particular test case, "a*" would expect to match "a" only, not the empty string as in "abab" (matching "a", then the empty string between the first "a" and "bab". String.split() expects the RegExp to be partially greedy.

This should be a documented deviation from ECMA-262 IMHO.

Comment 6

[Steven Johnson]

Any fix for this bug will likely need version-checking for backwards compatibility.

Comment 7

[Brent Baker]

Reset in-testsuite to '?' as this will require version testing

Comment 8

[Trevor Baker]

we bug is assigned to Serrano.

Comment 9

[Trevor Baker]

WE bug targeted to Brannan

Comment 10

[Sylvestre Ledru [:Sylvestre]]

Tamarin is a dead project now. Mass WONTFIX.

Comment 11

[Sylvestre Ledru [:Sylvestre]]

Tamarin isn't maintained anymore. WONTFIX remaining bugs.