Re: FYI: Making Atom Happen

[Steve Jenson <sjenson@xxxxxxxxxx>]

On Tue Dec 02 16:59:19 PST 2003, Dare Obasanjo <kpako@xxxxxxxxx> wrote:
 
> > How much poking is required to find giant holes in
> > RSS? Not
> > very many.
> 
> Give me examples, I'm an aggregator author so I'm sure
> I should be familiar with them.

RSS has the following problems:

1) What does link mean? What does the link point to? The original
content? Something the content is talking about?
2) What can description hold? Is it HTML? Is it single-encoded? Is it
double-encoded?
3) What does description mean? Is is the content of the page that
links points to or is it commentary on link?
4) What can title hold? HTML? Single Encoded? Double Encoded?

That's 4 giant problems with only 3 tags. And those are only the required tags.

So much developer time has been wasted because these things are not clear.
 
> > When Atom was proposed, RSS was cemented as
> > Finished by it's author. There was a Roadmap for
> > respectful innovation given and Atom followed that.
> 
> That was then this is now. The RSS spec is under a CC
> licence and anyone can publish their own version
> derived from it. What technical reasons are there for
> not doing so?
 
That's great but it's now too little, too late, I feel. It's 
nice that the author threw us some scraps, though.

If I were to start putting xhtml, without escaping it, directly 
into RSS's <description>, how many readers would throw 
up? If the number is any, then re-using RSS, using backwards 
compatibility as an apron to hide behind, is a dead end.

If I use explicit namespaces in my rss file (like 
<rss:rss xmlns:rss="xxx">(stuff here)</rss:rss>), then there are readers 
out there right now that will break even though aggregator authors are 
supposed to  "know better". I know this because I naively thought that 
RSS was really XML and I had Blogger generate lots of feeds that 
looked this way.

Those are bad readers, yes, but they aren't out of spec. That's my point.


It seems to me that part of Atom is helping move real XML, and 
not merely a format that looks like XML, forward in the syndication 
space. Real XML in the syndication space is a huge win. A mish-mash 
of kinda-maybe-mostly supported nonsense is seriously selling 
ourselves short. I'm ready to move forward.



-steve