- From: David Carlisle <davidc@nag.co.uk>
- Date: Mon, 29 Nov 2010 22:59:51 +0000
- To: Ian Hickson <ian@hixie.ch>
- CC: "public-html-testsuite@w3.org" <public-html-testsuite@w3.org>
On 29/11/2010 22:10, Ian Hickson wrote: > I don't recall off-hand if MathML has any specific requirements > that apply to data analysis tools but not Web browsers (or vice versa). In yes we have separate conformance requirements for mathml producing and consuming tools, and similar to html, rendering is formally not required but the test suite tests rendering in some depth anyway as otherwise things are just too strange. But the mathml test suite rendering tests are not really suitable here they consist of somewhat subjective visual tests against reference images. but the intention here wasn't to dump the entire mathml test suite on to the html test suite, but just to add some basic parsing tests (relating mostly to bugs fixed recently in various systems) plus just some basic rendering functionality tests, eg the one which is submitted which a system passes if it does more or less anything with mfrac different to span. I think the html5 test suite should be able to distinguish between systems that do and do not render mathml, I have no problem with having the mathml and svg rendering tests flagged in some way that makes it clear that they are testing foreign content and so perhaps treated differently in terms of html5 CR exit criterion, and we are happy to make the tests, just looking for input in what form they should take and where they should be checked in. If the structure isn't yet clear we can delay making the tests, or we can check in some anyway with the understanding that we'll change them again if needed. I just checked in these three to see what the consensus is.... David
Received on Monday, 29 November 2010 23:00:19 UTC