W3C home > Mailing lists > Public > public-html-testsuite@w3.org > November 2010

Re: choice of javascript DOM methods suitable for use in tests

From: David Carlisle <davidc@nag.co.uk>
Date: Mon, 29 Nov 2010 10:03:48 +0000
Message-ID: <4CF37A84.6000901@nag.co.uk>
To: Geoffrey Sneddon <gsneddon@opera.com>
Cc: "public-html-testsuite@w3.org" <public-html-testsuite@w3.org>
On 29/11/2010 06:47, Geoffrey Sneddon wrote:

> My gut reaction is testing stuff like this is undesirable, and that just
> using a negative reftest would be a better fit for such tests.

Well, as I said, this was just a small example to see what people 
thought before we committed the work into making more tests.

 > However,
> in this specific case, I'd go further and ask is there actually anywhere
> in the spec that expects MathML fragments to have any specific
> rendering? I can't quickly looking find one, and if not, it seems like
> such a test is out of scope.


True, although the same is true of the svg tests which have already been 
approved.

Also for HTML elements the first line of section 12 is

"User agents are not required to present HTML documents in any 
particular way."

So there are no normative rendering requirements on any part of html, 
however user expectation would be, I believe, that a system claiming 
100% test suite conformance did in fact render HTML5, including SVG and 
MathML.



Certainly there should be a separation between the parsing tests 
(clearly mandatory, for all media) and the rendering tests (clearly not 
mandatory, but strongly suggested). Currently with so few tests approved 
it's a bit hard to see what the planned overall structure is, and where 
each type of test should go.


>
>> Note the (current) intention here is to keep the mathml-in-html tests to
>> purely automatic javascript tests using the provided test harness.
>
> I'm not sure that's really a good idea for purely visual tests, FWIW.
> When we have a test type for visual tests, it seems better to use that.
>

I'd note (as a newcomer to this list) it's not exactly clear from the 
existing test suite documentation what test types you _do_have. There is 
an example test using a testharness javascript library, which I 
attempted to follow, and there are several other tests using several 
other styles, but no instructions on when of if those other styles 
should be used. I know more or less what a "negative reftest" is from
hanging around various lists but are there any instructions anywhere on 
how to submit such a test to the testsuite?

I can't see anything in

http://www.w3.org/html/wg/wiki/Testing

I assume you mean something like

https://developer.mozilla.org/en/Creating_reftest-based_unit_tests

My understanding of a reftest is it works best if there is a reference 
expression that should render in exactly the same way as the expression 
being tested. This may be hard to achieve for MathML. That's why I 
thought of using DOM access and the supplied testharness to just check 
some very basic details of rendering.

David





________________________________________________________________________
The Numerical Algorithms Group Ltd is a company registered in England
and Wales with company number 1249803. The registered office is:
Wilkinson House, Jordan Hill Road, Oxford OX2 8DR, United Kingdom.

This e-mail has been scanned for all viruses by Star. The service is
powered by MessageLabs. 
________________________________________________________________________
Received on Monday, 29 November 2010 10:04:19 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 15:49:37 UTC