- From: Andreas Strotmann <Strotmann@rrz.uni-koeln.de>
- Date: Fri, 25 Jul 2003 16:45:18 +0200
- To: Max Froumentin <mf@w3.org>
- CC: www-math@w3.org
Hello Max, Max Froumentin wrote: >Hi Andreas, > >Andreas Strotmann <Strotmann@rrz.uni-koeln.de> wrote: > > >>I hope this can be fixed along with the publication process for the >>second edition. >> >> > >The test suite does not claim to define MathML conformance, the spec >does. Therefore it is OK that it includes deprecated markup. It does >not mean that a conformant processor must support all the tests that >include it. The tests you list can be viewed as error cases that an >implementation must detect (like our XML Schema raises an error when >it tries to validate them). > But it does purport to *test* conformance (if not define it), so that errors should only be raised in the appropriate section (namely, error handling). Besides, if the MathML processor you're testing is in fact an off-the-shelf Mozilla browser, which (correctly) refuses to process the simple formula "a<b", then I question the appropriateness of leaving the tests as they are. >Besides, one could try to run the test suite on a MathML 1.x processor >where those tests should not return an error. > In which case it should be labeled as a MathML 1.x test suite, not "the" MathML one. As long as it is prominently linked from the main MathML page as *the* MathML test suite, its positive examples should not use any features deprecated at the current Recommendation version level at all, since it should serve as an example to other users of MathML. > >Cheers, > >Max. > Regards, -- Andreas
Received on Friday, 25 July 2003 10:45:23 UTC