Re: propose to approve tests - block 1

Chimezie Ogbuji wrote:
> Just as a heads-up the tests that GRDDL.py currently does not pass are
> indicated below.  A few of the tests needed to be marked as networked
> tests as they rely on .htaccess behavior.  I hope to try to determine
> why these are failing by tomorrow's telecon.
> 
> On Tue, 2007-04-24 at 17:13 +0100, Jeremy Carroll wrote:
>> #sq1
> 
> Still broken, but I'm assuming our outstanding modification to include
> all documents with /rdf:RDF in the root as RDF/XML documents will
> address this.
> 
>> #hcard1
>> #xhtmlWithGrddlEnabledProfileAndADisabledInBodyTransform
>> #xhtmlWithGrddlEnabledProfileAndInBodyTransform
> 
> I had previous issues with this one, see:
> http://lists.w3.org/Archives/Public/public-grddl-wg/2007Apr/0025.html
> 
>> #xhtmlWithMoreThanOneGrddlTransformation
>> #xhtmlWithMoreThanOneProfile
>> #card5n

Those six are a bit worrying.


>> #embedded-rdf4

this should work - my code failed at some point because of the following 
bug (i think):
- I had only one record of the current base URI
- reading the xml:base in the root element, I changed that record
- I then read the RDF/XML output by the transform with that base, rather 
than the retrieval URI, and hence invoked the xml:base attribute twice.

well it went away when I changed to have two records of the baseIRI and 
the retrievalIRI.


>> #embedded-rdf7
>> #embedded-rdf7-alt
> 
> I imagine the failure might have to do with the XSLT 1.0 fallback
One of these should work, but we might need to review whether using XSLT 
1.0 fallback is appropriate. The xslt 1.0 version (I think it's 
#embedded-rdf7) does assume that you ignore grddl results if the 
transform explodes on you.
We perhaps could mark the feature of the library as having an XSLT 2.0 
transform with XSLT 1.0 fallback as at risk.

> 
>> #grddlProfileBase4
>> #htmlbase3
>> #htmlbase4
>> #htmlbase1
>> #htmlbase2
>> #xmlbase1
>> #xmlbase2
>> #xmlbase3
>> #xmlbase4
> 
> These are probably due to GRDDL.py not yet conforming to our still-to-be-determined consensus on xml:base interaction with HTML Base as well as lack of support for the
> @@appropriate RFC for URI resolution@@ mechanism for using the *last* URI after a redirect (I hope to implement that before tomorrow's telecon).

Yes - although I deliberately did not add a test mixing xml:base and 
html:base, since xml:base is not permitted in HTML, and we have no error 
tests.



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

Received on Wednesday, 25 April 2007 12:38:10 UTC