Format Wars, Episode 3
I think it's funny when people say that feed formats are really irrelevant because all of the readers support all of the formats anyway. While that happens to be mostly true right now, the new, non-backwards compatible Atom Format spec is going to be released one of these days, and we'll be right back to the chicken-and-egg problem of publishers not wanting to convert their feed to the new version until it's widely supported in the readers, and the feed readers having to prioritize adding support for the new format and managing a release.
I think SmartFeed is going to be a big help in helping publishers manage that transition. When the spec is released, we'll make sure that SmartFeed "down converts" the new spec to the existing 0.3 spec for readers that haven't been upgraded yet.
What's in store for Format Wars, Episode 4 you think? I'm thinking it's going to be in reconciling different namespace extensions for rich media types. Different music namespaces, photo namespaces, video namespaces ... and how about enclosures vs. Y! RSS Media extension for the next generation of podcasting? All coming soon.
Comments
An interesting sidebar is that the new Atom format doesn't have a version attribute and I understand it will sit in the same namespace. I tried a few rss clients with the new format and many crashed.
Posted by: Randy Charles Morin | April 26, 2005 12:02 PM
I think calling them "Format Wars" unappreciated and negative. You're veering into territory where you're telling people what sorts of documents they can put on their servers.
Randy, you're just wrong on all counts.
Posted by: Robert Sayre | April 26, 2005 10:46 PM
Robert, as usual, I'm wrong and you don't tell me why. This is the most frustrating part of dealing with Atom. The WG evokes an attitude of superiority.
"Randy you are wrong and I'm not telling you why because you are too stupid to occupy my time."
Whatever. Boy, I'm in a horrible mood today ;) I was delayed in traffic for an hour.
Posted by: Randy Charles Morin | April 27, 2005 11:01 AM
Randy, no fair! You are repurposing RDF superiority arguments towards ATOMPUB! :)
You are correct that the format does not have a version attribute. However, the IETF-approved format does not and will not sit in the same namespace. If you read the rules in section 6, you will see that adding a version attribute at some point in the future will not trip up a conformant Atom processor. The WG's thinking was that we could add a version attribute if we need to introduce backwards-compatible additions to the format. Until then, it's "YAGNI". If we need to break backwards compatibility, we can change the namespace.
Posted by: Robert Sayre | April 27, 2005 01:17 PM
Eric, it's clearly your game to create, encourage and speak about the thing you call "war". I don't thinks it exists now and that we will see appear in the next future.
Posted by: François | April 27, 2005 04:47 PM
To Robert and François, I really didn't mean the title to be denegrating ... the title was just supposed to be a riff on the Star Wars movies coming out, but I can see that it didn't really hit very well!
I actually have high hopes for Atom and, as I mentioned in the post, think it has a good chance of being the de facto syndication format in a year's time. This despite the often dysfunctional community surrounding the effort (HHOS)!
Also, I learned long ago to not disagree with Randy on the finer points of interpretting the specs when he nailed me on an Atom 0.3 namespace issue last year!
Posted by: Eric Lunt | April 27, 2005 08:51 PM
Robert, you said "you're wrong on all counts." Then you said "You are correct that the format does not have a version attribute."
Also, what about backwards compatibility with 0.3?
Posted by: Randy Charles Morin | April 27, 2005 10:15 PM
Randy: I'm just catching up a bit here, and haven't been involved at all in the Atom process... posted on my own blog about the versioning thing, as evidenced by the trackback above.
My understanding is that the upcoming v1.0 of Atom isn't intended to be compatible with v0.3-- which was mostly just an ad hoc "standard" anyway-- but it lays the groundwork for some specific compatibility rules for revisions from here on out.
No version attribute, but the namespace will be different. I quoted this bit of relevant verbage from the spec in the abovementioned blog post:
"The namespace here is a temporary one and will be changed when the IESG approves this document as a standard. At that time, the namespace will be drawn from W3C URI space. The choice of that namespace will be coordinated between the IETF and W3C through their respective liaisons."
Posted by: l.m.orchard | April 28, 2005 07:17 AM
Well, I'm quite disappointed that compatibility with 0.3 has been forsaken. It seems too many people are willing to break the Web for religion.
Posted by: Randy Charles Morin | April 28, 2005 11:54 AM
Note, of course, that there is no published spec for Atom at this time, only drafts. That a pre-draft was so heavily evangelized and implemented is unfortunate. When Atom is published, those extent pre-draft feeds will be strongly deprecated (think: more effort than Syndic8 puts into fixing feeds). When the feeds are mostly gone will be the sign for client developers to stop maintaining pre-draft code.
There's really nothing to be compatible with, that's like saying a final product has to be compatible with its half-implemented prototype.
Posted by: Ken MacLeod | April 29, 2005 12:52 PM
Isn't this exactly why the IETF asks folks to not build implementations on pre-release specs. We shouldn't be complaining about some broken necks when folks have dove into the pool before it was filled. Here is the quote from the top of the draft spec...
"Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."
Posted by: Ken Kolano | May 5, 2005 10:16 AM