Some library tests

I have added more library tests to the pendinglist, i.e.

http://www.w3.org/2001/sw/grddl-wg/td/pendinglist

The eight tests I have done so far, exercise the xml-attributes module, 
with respect to the problematic SVG-in-XHTML DTD, which permits both 
xml:base and html:base.

Part of my motivation was to try and judge whether the option to respect 
or ignore xml:base inside html should be set to true or false by default.

The eight input documents are essentially the same, but with three 
different possible base declarations, (giving the eight combinations).

I have only been able to test the docs in firefox, since other browsers 
at my disposal do not appear to support these documents. I would be 
interested in feedback from other browsers.

The idea is that the results from extract-xlink1 correspond to what 
firefox does (except that extract-xlink1, following the library code, 
does not support nested relative xml:base's, found in two of the tests).

The results from extract-xlink2 are found by ignoring the xml:base.

My current intent is to leave the default as to ignore xml:base inside 
HTML, but have fairly positive documentation indicating that the option 
to respect them is a positive option.

A behaviour of the library module that is not tested by these tests is 
that it barfs if there is a relative xml:base and a relative html:base. 
This is intended defensively against an idiom like:

<html xml:base="foo/">
   <head>
     <base href="foo/" />

where it is plausible that the user intent (for this invalid input), is 
that one but not both of the xml:base and the html:base is respected by 
the agent. I don't believe that any of the other cases produce such 
conflict.

I'll e-mail my to-do list for the library separately, and also a request 
for stylistic advice.

Jeremy




-- 
Hewlett-Packard Limited
registered Office: Cain Road, Bracknell, Berks RG12 1HN
Registered No: 690597 England

Received on Tuesday, 19 June 2007 13:27:00 UTC