Re: [syndication] locating syndication feeds

[Morbus Iff <morbus@disobey.com>]

I'll respond to the other posts on this topic before giving my own:

>  a) OCS [http://internetalchemy.org/ocs/index.html] already does much
>     this and more.  I've never particularly been a fan of OCS -- gobs
>     of folks find it confusing -- but it is already understood and used
>     by aggregators and readers.

I would tend to say "if people find it confusing, we're not moving forward
happily, we're moving forward entropically". I've never used OCS in code,
but just looking at it makes me shiver. The channel lists that it creates
are HUGE and its mostly all repetition repetition repetition.

I'm a firm believer in "code shui" - much like feng shui, code has to look
good to make people want to work with it. OCS makes me want to drop dead
like that granny on TV.

>  b) Why not use RSS itself to produce channel lists?  Self-referential
>     niftiness aside, it's already well understood by those wishing to
>     construct channel lists and leaves room for syndication of
>     feeds.  Of course it would take a tag or three in 0.9x, but 1.0
>     require nothing more than a simple module along the lines of what
>     you've already mocked up.

This wouldn't be that bad of an idea.
You'd be looking at adding that as a namespace?

Ok. Thoughts on your piece:

>If a site has a number of syndication feeds, a channel
>list should be identified for each separate feed.

I worry this doesn't scale well. What happens with a site that has a
thousand feeds? They suddenly need to make a thousand extra files that
represent these feeds? And then a matching webpage that links to all these
thousand feeds in two different formats (<link> and user viewable?).

><rss092:channel>
><rss1:channel>

I almost wish we could change those around to

   <channel:rss092>
   <channel:rss1>

to make them more parent/child friendly. Would the various formats be
hammered out as part of any sort of DTD, or could people start adding their
own?

   <rss092:channel>
   <rss1:channel>
   <csv:channel>
   <psv:channel>
   <scriptingNews:channel>

And so forth?

I worry about the Content-type integration. How does that work for
aggregators that use the browser for their own UI? We'd have to set the
Content-type to the browser itself - can Content-type's be specifically
redirected to URLs? In the case of AmphetaDesk, could I redirect a
Content-type to:

  http://127.0.0.1:8888/add_channellist,xml?data=$data

If that's possible, then it'd also allow NewsIsFree.com integration.

The channellist.xml has no way of knowing what itself is. Say my aggregator
downloads a channellist.xml, tries the various feeds and they all work
perfectly. Suddenly, one fails. It tries another. Finally, after giving up
on ALL the feeds located in channellist.xml, it has no other options. It'd
be nice if there was a tag within channellist.xml that contained the URL to
the channellist.xml - thus allowing the aggregator to get the newest
locations. This allows webmasters to carefully care about only one URL,
instead of dozens or hundreds. (Although, quite frankly, that's a pretty
lazy webmaster).

>It might be nice to do something similar with the channel list: have it at a
>predictable place relative to the current base href ("channellist.xml" in
>the current directory), or if not found there, then at "/channellist.xml",
>but also allow a link rel meta tag to override and point the user-agent

But my poor error logs!

-- 
      ICQ: 2927491      /      AOL: akaMorbus
   Yahoo: morbus_iff    /  Jabber: morbus@jabber.org
   morbus@disobey.com   /   http://www.disobey.com/