Raw

Where are the RSS feeds of the W3C mailing lists? Eh? Eh?

PS. Here http://www.w3.org/2002/09/Lists2RSS?ml=public-rdf-dawg (replace the last bit with the list of interest, title & link only).

I tried gmane, was only able to get 2 of their lists I wanted, plus 2 other lists: list.opml. I've also pasted into opml 32 of the Google-Atomized feeds from Usenet groups that appealed to me (it's been a few years since I used Usenet regularly, so have probably missed some goodies): newsgroups.opml Grr, just noticed I didn't add the Java ones.

My subscription list at bloglines is up to about 2300 now, and I finally remembered to get a dump: export.xml. I tried the opml2blogroll xslt I'd done for the Democrat Conference blogroll on it, didn't work. Nor did Morten's 2ocs stylesheet. I can't believe it possible to create an XML format that was actually anti-interop, but opml proves me wrong. Anyhow I did a bit of tweaking and got something working enough for my purposes: opml2blogroll.xsl. The output is based on the FOAFish Planet style of RDFBlogRoll, cut down to what was available in the opml. Here's the result of applying it to the kilofeedlist: roll.rdf.

The RDF Validator seems mostly happy, this has 9344 triples, one apparently spurious accurate error report:

Error: {W107} Bad URI < http://quickbooks_online_blog.typepad.com/blogmain/>: Host is not a well formed address![Line = 12502, Column = 81]

A job for later is doing something useful with the categorization found in the opml, e.g. <outline title="postmodernism">. (Right now I've just got these dumped in comments). I don't think it would be all that useful to extract a 'global' taxonomy directly, the meaning varies from agent/person to agent/person (and the syntax! - I've also seen <outline text="postmodernism"> used).

Essentially you've got a tree of labels associated with an agent/person, and then from that tree a path is associated with sets of resources. That'd be equivalent to an (ordered) list of labels associated with each individual node. This kind of local, person or agent-specific taxonomy appears in quite a few apps, I think this might be worth the effort to find a general solution - SKOS maybe offering a likely approach. One for later and/or the LazyWeb.

Oh yeah, while I'm about it, do the following sums make sense? I've made a lot of assumptions, but was just looking for some ballpark figures for a practical desktop aggregator.

1. Transfer rate is 64 kbs
2. Average feed size is 10kB
3. Feeds are updated once per hour

How many feeds can be kept up to date continuously?

1hr = 3600s

64000 b/s = 230400000 b/hr

10kB = 80kb = 80000b

230400000/80000 = 2880 feeds

Which is so close to my feedlist target of 2500 it must be wrong!

PS. Bonus points to anyone that can exress this in Janets

[Danny]