I'm answering this from an aggregator's point of view.

In article <20010823171041.2B0DB36EC@...>,
notify@... writes
>1. Radio creates RSS 0.92, and many syndicators require 0.91. The reasons? 0.91
>includes important fields such as &lt;title that are optional in 0.92.

Slightly irritating is that the XML gif on many Manila sites points at
the scripting new format, not the RSS format. So now I know that (say)
http://scriptingnews.userland.com/xml/scriptingNews2.xml
should be converted to
http://scriptingnews.userland.com/xml/rss.xml

I've had to make sure that my parser can handle RSS 0.91, 0.92, 1.0 I
don't really want to have to parse scriptingnews format as well.

>2. There's an existing niche of "headline syndicators" that are distributing
>content, mostly vis RSS 0.91. Their items typically include a &lt;title, a
>/*brief*/ and a &lt;link to the complete story. Content created by Radio tends
>to look pretty bad when processed by headline syndicators, if they accept it at
>all.
>
>3. One reason Radio-produced content looks bad is the lack of an explicit
>&lt;title.
>
>4. Another reason Radio-produced content looks bad is the habit of some authors
>(myself included) to publish the complete text within the rather than provide a
>brief (1 to 2 line) , and an explicit &lt;link to the complete article.

There's plenty of slightly broken or awkward RSS out there. As an
aggregator I have to cope with html in Titles and Descriptions, long
Descriptions, no Link, no Title, etc etc. But I get to decide how it's
displayed so I get to decide how I'm going to cope with it. eg
- Strip all html from Title
- Limit how much of the description is displayed, if at all.
- Strip some html tags from Description if I don't like them
- Synthesize a Title from the first 30 chars of Description, if it's
missing, after stripping html.
- Use the first link in description or the sites url for Link if it's
missing.
- etc etc.

Should I have to do this? Well probably not. But I have.

Link is mildly contentious. A lot of syndicators use it to point at the
page with the full text of which Description is an abstract. Webloggers
tend to use it to point at the first link in their text rather than the
archive of their text. I don't happen to agree with this, but that's
their choice.

>First, since I want to syndicate my weblogs via the headline syndicators (and
>because I like Radio in most other ways), I'm going to modify Radio to use
>explicit &lt;titles and &lt;links, and to generate an RSS 0.91 file. (Wish me
>luck!) My version control will be a mess when Userland updates Radio, but I
>think it's my best option for now. If anyone else wants it, I can post the
>modified code when I'm done.

I've been down that route and managed to get a passable version of 0.91
out of Radio. (Actually 0.91 and a half; long description containing
html) I may or may not still have the code...

>By creating RSS files that are inappropriate
>for headline syndicators (re the original idea of RDF channels) are we
polluting
>the RSS gene pool?

I think this is taking it a bit far. Aggregators have to go to some
extra trouble to work with Manila/Radio RSS and mix it with RSS from
other sources. But it's such a large generator of RSS that it's worth
it.

-----------------------------------------------------------------------

And now the bete noire. Can we please sort out the title=permanent ... >
code that has started turning up. I've stopped reading Doc Searles
weblog because of it.

eg this snippet from today's http://scriptingnews.userland.com/xml/rss.x
ml

<title>title="Permanent link to 'Microsoft's domination of the Web' in
archives."><img src="http://www.scripting.com/images/leftArrow.gif"
height="9" width="11" border="0"></title>
<link>http://scriptingnews.userland.com/backissues/2001/08/23#microsof
tsDominationOfTheWeb</link>
<description><b>Microsoft's domination of the Web<a name="microsoftsDo
minationOfTheWeb">&nbsp;</a> title="Permanent link to 'Microsoft's
domination of the Web' in archives."><img src="http://www.scripting.com/
images/leftArrow.gif" height="9" width="11"
border="0"></b></description>

It's broken, right?

--
Julian Bond email: julian_bond@...
CV/Resume: http://www.voidstar.com/cv/
WebLog: http://www.voidstar.com/
HomeURL: http://www.shockwav.demon.co.uk/
M: +44 (0)77 5907 2173 T: +44 (0)192 0412 433
ICQ:33679568 tag:So many words, so little time