On Tue, Jan 13, 2009 at 5:05 PM, Bill de hOra <[email protected]> wrote: > > Aristotle Pagaltzis wrote: >> >> * Martin Atkins <[email protected]> [2009-01-13 19:05]: >>> >>> What are you referring to in RFC4685? (Atom Threading >>> Extensions) >>> >>> The "replies" link relation? I'm not sure that it means what we >>> need for this use-case. >> >> No, the `in-reply-to` element. It doesn't sound like the link >> relation was I thinking of at first glance but the definitions >> are congruent. Even my proposal for how it could be usefully >> displayed is essentially that of a thread. It's kind of >> embarrassing that I didn't notice immediately. >> >> (Nearly no one supports 4685, of course. But then neither do they >> support the hypothetical link relation.) > > I've used threading for both comments and capturing private messaging. I > agree it can work here, but a qualifier similar to rel would be better to > cover off Martin's bookmark use case.
It's the same issue that's come up again and again about expressing hierarchy in Atom. Proposed rel values include "detail"/"master", "up"/"down", "parent"/"child", or using the threading extension (the closest thing we have right now for hierarchy). None of those fit just exactly right for this use case, and I think a new rel value (esp. for pointing from child to parent a la 'in-reply-to') would be extremely useful. I'd hope it could be somewhat generic, though (and paired with a parent-to-children rel a la "replies" as well) so that it might serve these other uses. As Aristotle suggested, it simply needs to be more specific/strong than "related." --peter keane > > Bill > >
