We’re getting real, real close to sending the Atom data-format draft off for general IETF review; the rest of the process can’t happen too fast for me, because there are two big problems that bite me every day that Atom will give the engineers the tools to fix.
Damn the Dupes! · Via Technorati and PubSub, I subscribe to a bunch of synthetic feeds based on various keywords searches and URL linkages. They are infested with duplicates. There’s something about some people’s feeds that cause certain entries to show up as new every time the feed is fetched. Secondly, there are all sorts of primary and secondary aggregators out there; anything of note out in the world of Java has a chance of showing multiple times on TheServerSide and Java.net and Artima and so on.
The effect is, when I glance at my newsreader and see that there 12 new items in the synthetic-feeds section, I know that the chances are that in fact I’ve seen them all before. This problem is already bad and it’ll get worse fast, as we get more and more different aggregators and syntho-feeds.
With Atom, every entry is required to have both a
universally-unique ID and a field called updated, the time-stamp
when the publisher thinks it last changed.
That means that if ever I see the same entry twice, that’s a bug!
There’s no
wiggle room, somebody’s breaking a rule and I can track it down and harass
them until they fix it.
How Do I Subscribe? · Recently I got plaintive email from Postmodern Sass asking “how do I subscribe to a blog?” When I explained about the orange XML icon and the cut-&-paste dance I got an irritated note about about geeks’ inability to understand how Real People (tm) think. She has a point.
Molly Holzschlag ranted about this one too, yesterday. She calls for standardizing a bunch of different things; but I’m sure she’d agree that what we need is a portable, standardized way so that you can have a link saying “here’s my feed” and when you click on it you’re subscribed, right then.
In a previous fragment I pointed to contributions on the subject from Dave Winer, John Robb, Phil Windley, and others.
Once again, Atom gives implementors what they need to make this problem go
away. Atom feeds can (and should) contain a field named self
that says “here’s my address”, so that if you have a copy of the feed, you
have everything you need to know to subscribe to it, which often isn’t the
case with pre-Atom feeds.
