On Sat, 30 Jul 2005, A. Pagaltzis wrote: > > * Justin Fletcher <[EMAIL PROTECTED]> [2005-07-30 23:25]: > > I'm quite happy with 'replies-source' because it does not > > indComparing to mail seems to be a reasonable thing to do - > > mail has a considerable amount of prior implementation and > > understanding from which concepts can be drawn. However, there > > is no equivilent in mail terms for the purpose of the > > 'replies-source'. > > Iâm -1 on using âreplies-sourceâ; I should have said so more > explicitly before. To me it is non-sensical, as it parses as âthe > source of replies,â which is the opposite relationship of what > itâs supposed to express. Itâs not where the replies come from, > itâs where the entries being replied to come from.
Hmm. I see what you mean. > I suppose the criticism on âthread-startâ and âreferencesâ is > valid⦠but I still think that âreferencesâ is linguistically the > closest to what we are trying to express. > > Maybe âreferences-source?â >From what I've seen of comments it is the terms which are used which are causing problems. This may be partly because of the confusion over the use to which the words are being put - 'replies' is already specified as being the place where the responses to this reply will appear. 'references' implies that there are plural preceding resources being referred to when there is only a single resource referenced. 'start' implies the top of a heirarchy, when actually it could be anywhere within that heirarchy to which this resource is a response. Might it make sense to refer to the documents explicitly using the terms 'parent' and 'child' ? Whilst the mail terminology makes sense, I would be pretty sure that people would understand the family-relationship this invokes. Current draft term Familial term replies-source parent in-reply-to parent-id replies children This would result in links such as ... <link rel="parent" type="application/atom+xml" href="http://other.example.com/"/> <link rel="parent-id" type="application/atom+xml" href="tag:other.example.com:23"/> <link rel="children" type="application/atom+xml" href="http://here.example.com/stuff.atom"/> or, as a non-Atom-specific example .. <link rel="parent" type="message/rfc822" href="news:alt.foo"/> <link rel="parent-id" type="message/rfc822" href="news:[EMAIL PROTECTED]"/> It could be argued that 'parent' and 'child' should really be 'parent-document' and 'child-document'. However, it can be seen from the above that 'news:alt.foo' doesn't actually refer to a document, but merely to the method of locating the identifier referenced by the 'parent-id' relation. Does that make any more sense than the three names we already have ? Or is the problem I see only in my own head ? -- Gerph <http://gerph.org/> ... It's never sure, it's never pure, it always hurts.