From: David H. <da...@da...> - 2008-02-18 18:18:33
|
> > > > Does that make sense? > > Yes, but it's too much work :(. yes, I agree. I was just trying to run through the ideas clearly. > I don't want to start a browser rework until we've: > 1) Moved to PHP 5. > 2) Rearchitected the test case system to allow mixins. > 3) Added fluent mocks > > So the question becomes, should we make the change right now? to be honest, I'm still not clear what the proposed change is, or rather what the desired result is? The parser will probably parse utf8 encoded or latin-1 encoded documents without adding the 'u' flag. That's because all of the symbols which it's matching are encoded identically in utf8 and latin-1. > People using funny regexes will have to explicitly convert them with the > utf8() function in their tests, or use chr() sequences. people using funny regexes in their tests can pass in the u flag if they want anyway can't they?: $this->assertWantedPattern('/foo/u') > If we don't get any objections, then we'll just try it in the RC1 and > see who complains. not sure exaclty what you're suggesting to try out? I'd certainly argue against adding the 'u' flag in the parser (until we can do proper encoding transformation). The other place where regular expression matching is happening, ie in assertWantedPattern(), people control themselves anyway. David |