Re: CSS2.1 Test Suite RC2 Build Posted

So, bunch of XML parse errors in MS tests:

On 01/10/10 20:26, Robert Stam wrote:
> The files:
>
> at-charset-061.xht

Encoded as Windows-1252, yet has no XML declaration (it does however 
have a meta@http-equiv='content-type' which is meaningless there).

Probably on the whole better to just convert the test to UTF-8, as I 
doubt it'll come out of the build system as Windows-1252, and use 
something like é (\xC3\xA9) with @charset "ISO-8859-1" and check it 
doesn't match é.

> control-characters-001.xht

This test contains entities for U+0000 to U+0020 (surely that should be 
U+0019, the final Cc character of that range?), U+007F, and U+0080 to 
U+009F: only U+0009, U+000A, and U+000D are allowed below U+0020 in XML 1.0.

Heck, in HTML5, U+0000 is converted to U+FFFD, and the entities in the 
range U+0080 to U+009F are mapped to the characters \x80 to \x9F map to 
in Windows-1252, so if converted to HTML leaving those entities intact 
(which the build system does) the test wouldn't be testing what it 
intends on doing so.

I think what should be done is split the test into two tests, the first 
nonHTML test which tests U+0009, U+000A, U+000D as well as all the other 
characters from U+0020 and above that are interesting; the second 
HTMLonly test (where the original source format will have to be HTML) 
testing U+0001 to U+001F (or U+0020, if that is intended), U+007F, as 
well as U+0080 to U+009F (which mustn't be entities for the above reason).

> white-space-control-characters-001.xht

This is trying to test U+0016 (Synchronous Idle) doesn't affect the 
whitespace processing model. Simplest fix would be using U+007F instead.

-- 
Geoffrey Sneddon — Opera Software
<http://gsnedders.com>
<http://opera.com>

Received on Saturday, 2 October 2010 00:10:11 UTC