At work, one of the things I’m working on is a better way for information to flow to people. Since the one thing I know they will have always running is their email client, we’ve developed a solution that uses RSS feeds and an aggregator that runs inside the email client.
This seems to work well and I’m pushing for more advancement in this area.
I was playing around with the new Atom feed and started wondering why it displayed the way it did. I talked with Yoshi at work and he and I looked at the XML a bit. I asked why the stylesheet wasn’t referenced. After thinking a bit, he explained that the style information was an attribute of just one type of content that could be syndicated via Atom. To call it out specifically was wrong since it didn’t apply to all types of content. He said there must be a way to point at the CSS in another way, but he hadn’t looked at the Atom spec much.
We looked at the Joi Ito post about including CSS info in a RSS feed. We weren’t sure if this was a ‘good thing’.
A bit later Yoshi emailed me the following:
After some research, it appears that this approach is invalid. I quote:
https://www.w3.org/TR/html4/struct/links.html#edef-LINK
“This element defines a link. Unlike A, it may only appear in the HEAD section of a document, although it may appear any number of times. Although LINK has no content, it conveys relationship information that may be rendered by user agents in a variety of ways (e.g., a tool-bar with a drop-down menu of links).”
When I got home, I logged onto #joiito and started asking a few questions. I was soundly beaten up for suggesting that CSS and syndication even belonged in the same sentence. I argued that the layout and design of information was important. Mark Pilgrim, who’s opinion I respect, pointed me to his thoughts about Styles in syndication. He’s against it. I understand his reasoning, but I think the need remains. He suggested that style attributes be used instead of full CSS.
After putting the kids to bed, talking to the wife, and watching and episode of Angels in America with her, I headed back online.
In the mean time, Joi’s post went white hot with comments on the topic and Mamamusings made a great post about the topic. Anil Dash, always one step ahead, was mentioning this a year ago.
It appears that I stumbled onto the issue just as it is ‘being revisited’.
People on both sides make valid points. It’s good to be surrounded byt smart people.
My take is simple. For syndication to be truly successful, there must be a good way for style information to travel with the syndicated content.
The need for style info is more than wanting blue text instead of black text. For Atom to be more than a weblog tool and really shine as an information transfer platform, it needs to be capable of fufilling a range of needs beyond text & links. IMHO, Jason Shellen has a good proposal.
Thanks for the bevy of information, I’ve done what I can to propogate some of it back to the center of the discussion.
CSS into RSS will go
Overnight, there have been postings about enabling RSS feeds to carry some of the design, as well as the content, of its originator. See Joi Ito and then: floating atoll, mamamusings and (from his archives) anil. And add to these:
CSS into RSS will go
Silly of me to think we were about to be reduced to a series of ‘and here’s another …’ Instead, we have ‘and now for something completely different’. Overnight, there have been postings about enabling RSS feeds to carry some
Syndication has been successful in the news industry for years now. NITF is the standard XML format, and NewsML is emerging as the new way of doing things. If you ask aggregators of syndicated content what they want, they will tell you they want the data with no markup or style information at all. They (one example is Yahoo!) try hard to present a consistent experience for the user, which I would argue enhances their ability to find and process the information better than presenting each source of content in a different style.
It seems the trend is moving more toward standards like WSRP, which simply identifies a standard set of style tags to use within the content, but will be interpreted by the end system with their own style information.
Huh?