On Tue, May 31, 2005 at 12:13:42AM +0200, Thomas Broyer wrote: > <atom:entry> > <atom:title>New York City History</atom:title> > <atom:link rel="alternate" > href="http://www.columbia.edu/cu/lweb/eguides/amerihist/nyc.html"; /> > > <atom:id>http://www.columbia.edu/cu/lweb/eguides/amerihist/nyc.html</atom:id> > <atom:updated>2005-05-30T00:00:00+02:00</atom:updated> > <atom:summary>... Harlem.NYC - A virtual tour and information on > businesses ... with historic photos of Columbia's own New York > neighborhood ... Internet Resources for the City's History. > ...</atom:summary> > </atom:entry> > <atom:entry> > <atom:title>Gotham Center for New York City History</atom:title> > <atom:link rel="alternate" href="http://www.gothamcenter.org/"; /> > <atom:id>http://www.gothamcenter.org/</atom:id> > <atom:updated>2005-05-30T00:00:00+02:00</atom:updated> > <atom:summary>... Submit Events Edit Your Submission. Main > Neighborhood Stories NYC History in the ... The Gotham Center for New > York City History is supported by The CUNY Graduate ...</atom:summary>
I'm going backwards and forwards on whether atom:entry/atom:[EMAIL PROTECTED]"alternate"]/@href is the right thing to put into atom:entry/atom:id. On the plus side it's simple and works in the basic case of syndication (eg: my search results appear in an A9 search column). On the negative side I'm wondering what will happen if I take two search feeds into a desktop aggregator - presumably the result will only appear once. However that may well be what is desired, which would kind of be a neutral side. It does limit things (at least by draft-ietf-atompub-format-08) in that if I aggregate two search feeds into one and they have search results for the same site, either I have to drop one of them or I have to rewrite all the atom:id values myself - and the latter is explicitly forbidden by 4.2.7 of atompub. I know there's been some related discussion on this that hasn't made its way into the I-D yet - I /think/ consensus was reached around PaceAtomIdDos, but I can't find a statement one way of another on the list archives. With something like that text in atompub, at least people are aware of the problem, although something explicit in an Atom+OpenSearch document is probably required to make sure everyone's aware of what would happen. I'm also not convinced that the semantics are quite right (surely the feed is a feed of search results, where a search result seems different to me conceptually than the website a search result refers to as the source of the match). This isn't nearly so important, though, and I'm quite willing to accept the shmershing together of the ideas. (Or even that I'm wrong.) James -- /--------------------------------------------------------------------------\ James Aylett xapian.org [EMAIL PROTECTED] uncertaintydivision.org
