Archives for: October 2007

Monday, October 29th 2007

Permalink 04:43:03 pm, Categories: Meeting summaries

Meeting Summary 15th october

Work on the grouping of resources document has gone on for the last few weeks with significant questions still remaining. Specifically, attempts to devise a generic way to encode the data required to query any look up service has proved difficult, complex, messy, horrible - choose your adjective. As a result, the WG resolved to change the approach. It will be possible to define a Resource Set in terms of the properties of the resources within in it (e.g. the set of resources that are in the French language) and we'll make it possible to define characteristics of the HTTP Response Headers or body content. However, any look up service will be described in a separate (linked) document that could be written in natural language, WSDL or some other format. The idea is that a look up service operated by an organisation that regularly creates Description Resources, such as trustmark operator, would be at a location repeatedly cited in its DRs. In some ways this is similar to the HTML Profile concept. Following further discussion of the use case suggested by Liddy Nevile, the existing Use Cases and Requirements will be updated. The new use case makes more explicit use of profile matching as well as making a clear distinction between first and third party labelling. The group members whose primary concern is using POWDER for highlighting resources that meet accessibility criteria are taking on this document revision. Attention then turned to the Description Resources document. First of all, the group will create a profile document that can be cited in an HTML HEAD tag that defines rel="powder". Our documentation and examples can then point to this. This working group will join others in arguing that HTML Profile should be retained by the HTML 5 Working Group and not deprecated. The forthcoming Technical Plenary has a session on HTML 5 and XHTML 2 and this seems like an ideal opportunity to raise it. Discussion then turned to the more difficult issue of the HTTP Link Header. Efforts will be made to prove interest in this issue, in particular, by encouraging the re-issue of Mark Nottingham,'s IETF draft on the subject (or something very much like it). By coincidence, it was being discussed by Mark Nottingham almost simultaneously in the W3C IETF/HTTP WG mailing list and by him and Dan Connolly on this group's public mailing list a little later so getting HTTP Link and HTTP Profile headers seems at least possible. However, the group decided that if it proves not to be so, our own suggestion of using HTTP Link headers will move to the (non-normative) primer document, scheduled for publication later this year. Overall, the group is hoping to have new versions of the Grouping of Resources and Description Resources documents in the public domain by the end of the month and for these to be very close to being completed. The group will review them carefully at the face to face meeting in Boston in November and aim to move them to Last Call immediately afterwards.
Phil ARCHER

Tuesday, October 16th 2007

Permalink 08:34:12 am, Categories: Meeting summaries

Meeting Summary 8 October 2007

The main topic of conversation this week was, again, the Grouping of Resources document. In particular, grouping my by property look up. The present published version of the document defines the includeConditional RDF property which states simply what property (properties) a resource must have to be a member of the Resource Set, without defining how to look it up. This is currently flagged as a feature at risk. The document goes on to suggest a method through which a look up system may be defined using simple HTTP requests. This section is going to be improved with a more well-defined syntax but is likely to remain more or less as is to handle a limited set of 'simple situations.' More complex look up systems using a Web Service require more thought! The feeling is that defining a completely generic structure in which any look up service can be encoded is probably not feasible. As a result the plan is to make a 'best effort' and then seek comment. The whole section on Resource Set membership through the properties is likely to be flagged as a feature at risk. The main reason for keeping it is that it should make it easier to publish DRs making use of existing database systems. There was some further discussion of which regular expression syntax to use. The issue has now come down to using either that defined by XML Schema or XPath functions. As we're only using REs for simple matching, only a simple syntax is required - so the issue is really whether the anchor characters should be explicit (XPath) or implicit (XML Schema). Next week's meeting will review the grouping of resources document again and will certainly aim to publish it, ideally as a Last Call. Finally, the group noted (and welcomed) the submission of a new use case by Liddy Nevile. The group's accessibility champion will review this carefully and advise the group on whether it should be added to the document or not.
Phil ARCHER