Re: [css3-color] bug in table-based test

L. David Baron wrote:
> On Wednesday 2007-10-24 15:35 -0400, fantasai wrote:
>> It's caused by using a generic XML parser without reading in the XHTML 
>> entity files.
>> I've written a patch to preprocess the files first (since getting the 
>> parser module
>> to read in the entities from a local file doesn't seem to be possible). Let 
>> me know
>> if it looks good and I'll check it in.
> 
> You're running the preprocessor over the copied files, not the
> original source files, right?

Right.

> I suppose this looks ok, although I tend to think we're better off
> keeping the preprocessing to a minimum and just fixing the tests to
> use   rather than   (which is what I was planning to do
> before your patch).  It can be dangerous to change the preprocessing
> for existing tests.
> 
> Did you diff the 2.1 test suite output before and after your changes
> to make sure the changes make sense?

Yes. Looked reasonable. I was surprised that we have tests using some
of the fixed-width spaces, but the results looked reasonable.

> I'm a little worried about
> some of the whitespace tests and things like that.  I suppose this
> shouldn't really change anything, but it might in some user agents,
> and there could be some tests testing both entities and raw
> characters.

If using named entities turns out to be a problem, we can switch it to
hex codes pretty easily. I think making these invisible characters visible
helps a lot in making the tests understandable, though.

> Some slightly more detailed (but not all that important) comments on
> the code below.
...
> Then again, you could probably simplify the whole thing by using
> "perl -pi" and just making preprocess.pl just act as a stream
> editor, which I think would mean it would just be one line of code.

Thanks a lot for the review. Modified patch attached:

~fantasai