Consider the following code with SM output: "123".indexOf("", 0) result: 0 "123".indexOf("", 2) result: 2 "123".indexOf("", 3) result: 3 "123".indexOf("", 4) result: 0 I think the last one should also return 3, like in Chrome, Safari and Opera.

Note that this also fails in Firefox 3.6.

This fixes the problem, passes tests and adds tests. brendan can you review or should i ask someone else?

Looks like a regression from bug 511777, the old code also used textlen in this case.

The "regression" I believe was actually a bugfix: 15.5.4.7 step 5 says that, if the start position is greater than the length of the string, it is clamped the length of the string, not set to 0. I wonder if the other three mimicked our old, incorrect behavior.

(In reply to comment #4) > The "regression" I believe was actually a bugfix: 15.5.4.7 step 5 says that, if > the start position is greater than the length of the string, it is clamped the > length of the string, not set to 0. Well the problem is that it's set to 0. This patch clamps it to the length of the string, so that it returns 3 for the example in comment 0.

Oh goodness, I read the patch backwards, excuse me ;)

Comment on attachment 491155 [details] [diff] [review] Patch Jan, could you please add this to your .hgrc? [diff] git = 1 showfunc = true unified = 8 It looks like you don't have showfunc = true at least. Patch is fine, thanks a lot -- good to have test coverage gap filler too. /be

(In reply to comment #7) > Jan, could you please add this to your .hgrc? Cool, done.

http://hg.mozilla.org/mozilla-central/rev/13f5b057c586 Please include the bug number in your patch comments in the future. I failed to catch this, so right now there is no way to link back from the landed patch to this bug. :(

(In reply to comment #9) > Please include the bug number in your patch comments in the future. I failed > to catch this, so right now there is no way to link back from the landed patch > to this bug. :( OK, will do that next time; thanks for committing this.