PaceLinkConstruct (was Re: AtomPubIssuesList)

[Danny Ayers <danny666@xxxxxxxxxxx>]

Sam Ruby wrote:

> [1]<http://www.intertwingly.net/wiki/pie/AtomPubIssuesList>
Excellent work, thanks Sam.

Re. PaceLinkConstruct, it may be stable in itself, but adopting this 
would preclude other suggestions - for a start it only lists 10 rel and 
3 rev values where LinkTagMeaning has 18 (including parent and related, 
under discussion at PaceLinkParent and PaceLinkRelated). The bit that 
bothers me most is the removal of the possibility for extension.

So I wonder if it might be possible to word the next spec draft in such 
a way that allowed the current interpretation but still left room for 
additional values (e.g. parent, related) and/or a namespace and/or 
full-URI based extensibility mechanism. In other words something like:

"If the value of the rel/rev attribute is not one of those listed below, 
then its interpretation is undefined and the entire element MAY be ignored."

[list]

Norm says on PaceLinkRelMechanism there will be QNames in attributes in 
the next HTML, so I would suggest the replacement of 
PaceLinkRelMechanism with a QName-based alternative. I don't really like 
them but they're better than nothing ;-) I doubt I'll have time myself 
in the next few days to do a new Pace (and am a little reluctant when 
there is another pending extension mechanism that may render the link 
approach redundant). Once something's up on the Wiki, there still needs 
to be discussion so being realistic it seems to make sense to postpone 
any final decision.

Cheers,
Danny.




--

Raw
http://dannyayers.com