proposal: e-mail consensus? on test change #htmlbase1, #htmlbase3 and library

This proposal is subject to being seconded by John and Chime.
If they choose to resolve the #xmlbase1 test one way (the way I was 
initially proposing) then this proposal is incorrect. This proposal 
patches the other files if that test is resolved on Chime's reading of 
RFC 3986 (which I think is slightly better than my own earlier reading).

I PROPOSE that we modify the test decisions of the last telecon in the 
following fashion:
- tests #htmlbase1 and #htmlbase3
    are approved with the output files currently in the test directory
    which differ from those that were approved.
- to approve today's substantive changes to both library files, 
impacting their behaviour in any test using those files, and an explicit 
base (e.g. the above two tests, and #embedded-rdf4, #embedded-rdf5, 
#embedded-rdf6, #embedded-rdf8)
   [Note: the input/output files for these impacted tests do not change,
    but a GRDDL aware agent that would have passed the approved tests
    might not pass with the change to the library files]

===

Technical note:
   the difference, discussed to death on the #xmlbase[1234] thread, is 
whether the 'retrieval URI' or the 'base URI' using the in-document base 
is used for processing the RDF/XML output of the transform.
This proposal uses the 'base URI' using an in-document base: either 
xml:base on the root element, or html:base in accordance with the 
HTML401 spec.

===

These changes are currently checked in; the old versions of these files 
are attached to my previous message.

===

I have discussed the process for this e-mail consensus vote with Harry, 
and suggest the following.

A) if John and Chime choose to second, then they should do so today.
B) if anyone objects to this decision, then consensus was not achieved, 
and the matter will need to be dealt with in a telecon
C) if by *time to be decided*, the proposal has been seconded by John 
and Chime, and no objections have been received, then Harry will 
announce 'So resolved'

I will let Harry decide on the appropriate time, I suggest there should 
be at least 24hrs to consider (possibly excluding the weekend).
I think that still leaves us on-track for Last Call publication.

Jeremy

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

Received on Friday, 27 April 2007 09:46:25 UTC