[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [syndication] shared feed lists
Bill Kearney <ml_yahoo@ideaspace.net> wrote:
For sites wanting to offer a list of RSS feeds, use of the HTML <head> section
<link> element should be considered. Examples of which can resemble:
<link rel='subscriptions' type='http://purl.org/ocs/directory/0.5/#'
href='http://example.com/list.rdf'/>
Some thoughts on all this.
1) I've got a very similar requirement for large numbers of foaf files.
eg <link rel="alternate" type="application/rdf+xml" title="FOAF Scutter
plan" href="http://www.ecademy.com/robotfoaf.rdf"; /> where robotfoaf.rdf
is currently 2333 lines long. We also produce large numbers of RSS files
so the need is clearly there. Note that the rel="alternate" is almost
certainly wrong in the example...
2) Having a spec that says <link> in the <head> of the html returned by
the root of a directory is more flexible than
thisStandardFileName.extension in the root of a directory. But we're
stuffing a lot of these into <head> now. Which one is the subscriptions
list? So now we have to allocate a rel= value to this specific need.
It's enough to make you think there should be another layer of
indirection here; have every web page include a single <link
rel="metadata" href=""> which points to a file with a list of metadata
files each of which can contain one or more bits and types of metadata.
I know "every problem in computing can be solved with another layer of
indirection", but this is getting a bit daft isn't it?
3) Let's say we use <link> in <head> of the <html> returned by the root
of the directory. Who's going to come up with the example of where a GET
http request to the root returns XML or RDF? eg
http://www.voidstar.com/foaf/ What then? Can we assume that
4) I don't much like OPML for this because using attributes instead of
sub-tags looks messy to me (There, I've said it). But then I'm not a big
fan of OCS either. On the surface this is a pretty simple requirement
and I'm not sure why I should need a full RDF parser to extract the
names and URLs of the feeds.
5) I don't see how we can get away with saying only RSS feeds should
appear in this list. Apart from identifying the different types of RSS,
you can pretty much guarantee that someone will want to include atom
feeds (or whatever its called now). I can see a need for listing off OCS
feeds as well. I can feel this heading into the conclusion of 2) above.
this here list file could be a general purpose Metadata pointer file.
Sorry, more problems than solutions.
My unconsidered opinion is that a way of listing RSS feeds in a machine
readable format is a good thing. So I agree with DW on this.
And putting a <link href> pointer to it in the <head> of the index
document in directories is a good thing. Even if there's *also* a
standard name for the file.
But then I'll probably change my mind tomorrow or at least think of more
issues that this throws up.
--
Julian Bond Email&MSM: julian.bond@voidstar.com
Webmaster: http://www.ecademy.com/
Personal WebLog: http://www.voidstar.com/
M: +44 (0)77 5907 2173 T: +44 (0)192 0412 433