[GROOVY-4392] asymmetry of closure parameter type among String#replaceAll/String#eachMatch/Pattern#each

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.7.6, 1.8-beta-3
• Component/s: groovy-jdk
• Environment: MacOS X
• Assignee: Guillaume Sauthier
• Reporter: UEHARA Junji
• Created: 05/Sep/10 09:25
• Updated: 15/Dec/10 16:56
• Resolved: 21/Sep/10 10:52

Description

The closure which passed to String#replaceAll() takes arguments in two patterns:

1) Object[]
2) Arguments correspond to matcher's group(0), group(1), group(2) ...

but the closure for Pattern#match(Closure) and String#eachMatch(Closure) takes:

1) List
2) Arguments correspond to matcher's group(0), group(1), group(2) ...

So there is an asymmetry about 1).

If this issue should be fixed, IMHO, to change replaceAll() to take a list like
match/eachMatch is better. Because you can write simply:

assert "ABDE" == "abcdef".replaceAll(/(..)(.)/) { it[1].toUpperCase() } // (A)

rather than:

assert "ABDE" == "abcdef".replaceAll(/(..)(.)/) { Object[] it -> it[1].toUpperCase() } //(B)

(A) is same as:

assert "ABDE" == "abcdef".replaceAll(/(..)(.)/) { List it -> it[1].toUpperCase() } // (C)

I attach small patch for 1.8-trunk and some test cases for this.
This patch leads small backward-incompatibility so that (B) would fail because
of the difference of parameter type of the closure. But current test cases
don't include this.