On Mon, Dec 20, 2004 at 01:59:57PM +0200, Markus Laire wrote: > > I was wondering if it would make sense to add the original 're_tests' > file to parrot distribution, with a script which autogenerates > 're_tests.t' from it. This way it would be possible to update the script > if testing-format is changed, or if some bigger mistakes are seen. > > Of course small errors in 're_tests.t' file could be fixed manually, but > if testing-format it changed, then those changes would be lost when file > was autogenerated again.
After thinking about this a bit... Although it's probably worthwhile for us to find a good way to encode p6rule tests in a way that allows us to go to other formats in the future, I'm not so sure we buy a lot in being able to do that for the original 're_tests'. As I'll note in my next message, the semantics of regexs and rules are just different enough that I don't think that a one-to-one correspondence for 're_tests' is worth trying to preserve or code around. At some point I think it's just better to use the autogenerating script to create re_tests.t, manually fix re_tests.t for the things that don't convert, and then go from there. Plus, I suspect that p5's 're_tests' is mature enough that it doesn't change often, and the places where it might change in the future are esoteric enough that it'd be easier to add+translate those changes manually than to update the conversion script. Note that for this I'm only talking about ease of building and maintaining the p6rules test suite. Having a script to automatically convert any arbitrary p5 regex to its p6 equivalent *is* important, useful, and definitely worth pursuing, but I'm not sure it's worth the effort (at this stage) to worry about keeping the p6rules test suite consistent with that. And the p5->p6 regex converter will probably deserves its own test suite, once p6rules is sufficiently advanced. Pm