| Wednesday, June 07, 2000
Sun supports SOAP
News.Com: Sun changes tune in support of SOAP protocol.
"Sun executives say they decided to support the technology after looking at the newest version of SOAP, which included development work from IBM and Lotus Development.
"Sun is the second company to flip-flop on SOAP. In April, IBM also reversed its stance, saying the need for the software industry to find a way for businesses to link their different computing systems outweighed competitive issues."
What a total surprise, and welcome news.
MSNBC: Judge orders Microsoft breakup.
Other reports: CNN, Seattle Times, Washington Post, NY Times, News.Com, ZDNet.
Press release: Microsoft Confident That Judicial System Will Overturn District Court Ruling.
Seattle on Tuesday
Following on the success of the Scripting News dinner in Amsterdam in May, let's do the same in Seattle next Tuesday.
I'm visiting Microsoft, I have an 11AM meeting to get briefed on the "orchestration" technology; I'm told that the XML format will be open and submitted to a standard's body. Should be really interesting.
And it'll be great to meet with Seattle-based Scripting News readers. These dinners are fun. Brent is going to pick a restaurant, probably in downtown Seattle. Shall we start at 7PM?
Survey: Coming to dinner on Tues in Seattle?
Now, let the flames begin.
There's always a resistance to movement in any online community. It's been a long time since RSS has moved. Now I want to get prepared for movement. This is a total exercise in politics and group psychology.
RSS 0.91 was a major traffic accident that turned out pretty well. Netscape did a private spec, just for their own web service. I had a couple of problems with it. First, we had been active in the area, and there had been no attempt to move compatibly. Second, they had a publishing agreement that was really offensive. We dealt with the first issue by competing. They quickly saw the advantage of working with us. Hat's off to Netscape. And we routed around their agreement by opening My.UserLand, which in no way attempts to control the content that flows through it.
After RSS 0.91, we breathed a sigh of relief that lasted almost a year. Glad that's over! In the meantime, repeated attempts to find anyone who cares about RSS at Netscape have turned up nothing. The people we worked with at Netscape left shortly after 0.91 was finalized.
At WWW9 an impromptu BOF for RSS happened at Dale Dougherty's Web Publishing session. There were a fair number of people there who were actively developing in RSS. Some surprises, I didn't know that Sean McGrath was doing RSS stuff. I came out of the meeting with a resolve to get RSS moving again.
Resistence to movement
There's always a resistence to movement. It stirs people's feelings of powerlessness, at least that's how I parse it. Why isn't my pov being considered? is the constant question, sometimes not asked so nicely.
I've learned that it's necessary to strike a balance with this and the opposite complaint. "Why aren't you addressing my needs?" This is the question that comes from content providers. The answer is basically, if I move they're going to yell at me! (The other people, the ones who feel that their pov isn't being represented.) Then hopefully there's a point where the frustration over lack of movement overcomes the frustration of being powerless, and that's when movement can happen. I hope that we're now at that point.
Another observation. Every time a mail list starts to guide the evolution of a spec, it seems to generate more "Stop Now!" energy than "Let's go forward" energy. The two recent exceptions in my experience are RSS and XML-RPC. My philosophy on XML-RPC, based on the frustration in previous projects, was "I want to do it the way you want to do it." So when Don, Bob or Mohsen said "I think we should call that element flebangolaleo," I didn't say a word, I just changed it in the spec. That lead to a collegiality that I've never seen in a cross-company development project before. It was truly magical, imho.
Now the question is, can we do that with RSS? Can we avoid arguments over small issues, and agree on some basic enhancements that give content developers enough room to do the kind of innovation they want to do? That's the first question on my mind. And I come to it not only as a developer of a CMS and an aggregator, but I am also a content developer. RSS has not grown with my needs. The same is true for Motley Fool, I assume (if they're still paying attention) and many other intelligent content people.
(BTW, a hat-tip to Zeldman, who finds the term "content" derogatory. Me too, I've even complained about it publicly. But in these conversations, we need a convenient short-hand to describe the wide variety of creative people who combine to create the stuff that people read on the Web. Actually the "content person" we have contact with is often an engineer who writes scripts to adjust the content to fit the format the aggregators can consume.)
Can we keep it simple?
The second question, which to me isn't much less important, is Can we keep it simple?
Today RSS *is* simple, largely because it only builds on XML 1.0, and does not use namespaces or schemas, and it isn't a dialect of RDF. There's a logical route forward for RSS that says it should adapt to include all these concepts, but in doing so it would become vastly more complex, and imho, at the content provider level, would buy us almost nothing for the added complexity.
Further, imho, the use of namespaces punts on the larger issue, what are the names of common elements for channels and news items? Using namespaces just pushes the problem into a corner. Instead we could agree that the names we use are going to be imperfect and misleading. But there will be standard element names for the info that the content providers want to publish, in addition to the elements that RSS already specifies. (A canonical example is Motley Fool's desire to include ticker symbols with each item, a totally legitimate desire, imho.)
A place to start
So over the weekend I started a review of RSS 0.91, and took notes, looked at the actual XML that people are publishing, studied the Netscape specification, and created my own version of the 0.91 spec.
Up until this morning I wasn't sure if this document should be called 0.91 or 0.92. I was concerned that practice had deviated from the Netscape spec, esp in respecting the limits it imposes, which most developers (myself included) think are ridiculous and unweblike. But I found that most of the channels are respecting the limits, so I changed the title from 0.92 to 0.91.
So all this is is a cleanup. All the Netscapeisms are removed. It's better organized and easier to follow. It links to helpful W3C specs and IETF RFCs that provide the foundation RSS builds on. It has a timeline that points to earlier specs, and to three important aggregators.
I do not want to call today's version of the spec final in any way. So for now it has a simple copyright notice, but I expect to finish my work on this by the end of the week, considering that it doesn't add anything to RSS, it should simply be a matter of reviewing it for editorial mistakes or omissions, there is no room to debate new features, because the spec doesn't attempt to add any.
So with all those caveats and preambles, here's my version of the RSS 0.91 specification.
Dave Winer: RSS 0.91.
Comments, questions and suggestions are welcome.
Frontier 6.2b16 was released overnight. Two long-standing UI issues were addressed. And the Mac version's TCP functionality (which is rewritten in 6.2) gets two important fixes.
Luke Tymowski on Matt's Frontier book: "I read bits and pieces of it often, even though I have neither a Mac nor Frontier. Why? It's the best written computer book I've ever come across, worth reading for the writing alone." Wow!
Surprise: My car is my castle. "Probably the tiniest car is the Italian Cinquecento. It is this small that two to four men can lift it and carry it somewhere else. It fits in the smallest parking spot and is ideal for driving in the city. You can react so quickly it is like driving in a glove."
A smart car in Amsterdam, which is also pretty small.
NY Times: "We've been fortunate with our own alliances," he said. "But it's absolutely true that many of the ones you read about are ill-thought-through, and are done for the P.R. value. If there's one great thing about alliances, though, it's that they can be undone quickly."
Open Source is on the mind today. Talking with Edd Dumbill over the weekend, about lots of things, including the evolution of RSS, and the O'Reilly Open Source Convention, which was briefly a topic on Scripting News on Sunday.
I said something to Edd that I have not yet said on Scripting News, and if Tim had been willing to engage in a conversation, he might have said it himself. (To be clear, what follows is Dave in Tim's body.)
"Dave, get a clue. When we started the open source marketing push, followed by the convention, there was no such thing as the open source community. None of the developers or users of any of the scripting systems (Perl, Python, Tcl) talked to each other. Same with the developers of Apache, Sendmail, and MySQL. Now I understand your complaint that there are silly barriers that keep you from working with people in the open source world, and I wish there was something I could do about it, but look at all the good we've done. Now at least once a year they all meet in one building. This has lead to some wonderful cross-pollination, for example the recent decision by Zope to embrace Perl. These communities tend to build really large walls around themselves and tend to see the whole world as being inside those walls. We make progress as fast as we can."
Now, back in my own body. "Tim, thanks for pointing that out."
Edd also made a point that I've heard echoed by others at O'Reilly. Not all of the people in the open source world have embraced the Web. There are so many Venn Diagrams you can draw. Open source or not is just one of the diagrams, one set of walls that define communities. Another is "Are you a Web person or not?" I remember the look on Tim's face at the Web Apps session at Esther's. He was delighted. He asked "Why isn't the whole show like this?" An insightful question. Now back at you Tim, why isn't that point of view represented at the Open Source Convention?
I guess the bottom line for me is this. Are running a conference to make money and achieve a modest level of cooperation within the open source community, or are you running a Web conference, to draw more people together, to provide a way for people in ever more disparate worlds to get to know each other and work productively?
At the end of the Web Apps session at Esther's, Tim complimented me on the job I did. He asked if I was going to do this again. I said YES! He asked if I would do it with O'Reilly or on my own. I said at O'Reilly. Well they seem to have forgotten that, so we must do our own conference, and of course we will welcome people from the open source world. It's the old route-around thing that the Internet does so well. Be inclusive. Exclusivity is inevitable, but when it's discovered, treat it as a bug, and fix it, cheerfully. Murphy's Law. Not a big deal. "It's even worse than it appears."
Jai, Matt and the bathroom (And Visio)
A letter I sent to Jai Singh, editor of News.Com, asking about their policy re including UserLand in news stories.
Matt Haughey: "I am truly fucked."
Seth Gordon: "The Visio user interface sucks."