W3C home > Mailing lists > Public > public-css-testsuite@w3.org > November 2004

Re: Minor spelling

From: Tantek Çelik <tantek@cs.stanford.edu>
Date: Mon, 22 Nov 2004 20:56:42 -0800
To: Ian Hickson <ian@hixie.ch>, "Justin Wood (Callek)" <116057@bacon.qcc.mass.edu>
CC: Michael Day <mikeday@yeslogic.com>, <public-css-testsuite@w3.org>
Message-ID: <BDC802FD.50ABB%tantek@cs.stanford.edu>

On 11/22/04 7:58 AM, "Ian Hickson" <ian@hixie.ch> wrote:

> On Sun, 21 Nov 2004, Justin Wood (Callek) wrote:
>>>> 
>>>> Well, they're not testing PNG support either, but sure.
>>>> 
>>>> I've converted all the GIFs to PNGs. This will probably cause IE6 to
>>>> fail on a couple more tests.
>>> 
>>> This is contrary to some of the CSS test suite philosophy, which is to
>>> not test the things that are not being tested.  And given that GIF
>>> support is more prevalent than PNG support, of the two, GIFs should be
>>> preferred.
>> 
>> In that case I would propose and request that a version of the same test
>> be released both with PNG and GIF versions, with a note that EITHER
>> version's success will "PASS"...
>> 
>> Yes GIF is more prevelant but current Copyright issues prevent some UA's
>> from supporting GIF at all.
> 
> Patent issues.
> 
> But: which UAs do not support GIFs yet do support enough of (X)HTML and
> CSS that they will be relevant when showing interoperability of CSS2.1?
> 
> (Note that our exit criteria excludes UAs that are not "intended for a
> wide audience" and that could not "be used on a daily basis".)
> 
> The simplest compromise from my point of view would be to make the 1x1
> transparent image (the only image that is likely to cause WinIE any
> trouble) a GIF, and the rest PNGs.

I believe that should work.  Could you create a "test setup" page with the
sample images so that we can try this out?

> Or to make the 1x1 image both GIF and
> PNG and use HTTP content negotiation to serve one or the other depending
> on what the UA claims to support.

That should also work, yet borders on violating the methodology of sending
the same test to all browsers (the methodology which is thoroughly violated
by the DOM Test Suite thus invalidating its usefulness whatsoever for
testing interoperability).

> (But bugs in WinIE's HTTP implementation
> may make that self-defeating.)

I don't believe so.  Nonetheless that is just conjecture until we try out a
"test setup" page.

Tantek
Received on Tuesday, 23 November 2004 04:56:46 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 20 September 2010 17:51:54 GMT