Re-learning old tricks
Terry Teague reports that his disk crashed this week too, and Sylvia Paull says she had a bike crash on my birthday somewhere in Marin, and turned it into a business development opportunity. I'm still enjoying the clean life of a person with little or no past, hard-disk-wise. Today's the day we're supposed to get the results from the repair service. As Dionne Warwick sang "Say a little prayer for me."
Good news. The hard disk repair company was able to recover the data. I should get the most important bits back within the hour and all of it within the next few days. Thanks for all your prayers, it worked!
And that was just the beginning. I had lost all my favorites in Radio, and all the little scripts I had written that made things automatic. Some of our services broke and there was nothing I could do about it until the data came back. If it hadn't I would have had to start over. When the data finally came back, I felt a physical release. I had to go collapse. I slept for three hours. Not a restful sleep. My subconcious thinks that some of those bits are necessary to my survival. Unbeknownst to me my life was hanging in the balance, according to my subconscious. I took a walk and did a lot of breathing. I asked "What would make you feel safer?" A simple answer. Better backups.
Now I have a couple of fewer days to get my new software done. It's got a codename and a plan. Its codename is Smurf Turf. I'll explain later.
Priorities for XML-RPC and SOAP
As we started making progress on the interop testing in SOAP and then XML-RPC, I started working on a list of new priorities, for the post-interop period (which I hope lasts for years). Today I added a new priority, deployment. But it seems a good time to review them all: interop, apps, deployment, business development.
1. Interop is a process that goes on forever. There will always be new toolkits coming online. However, with every passing month interop will get easier. Related to interop, I plan to do a revision of the BDG, perhaps Jake can do the work, to remove the concept of a "subset" from the document. It's finally sunk in that it's not a subset. We've had continuing talks about the philosophy of interop, esp as it relates to what interop means in HTTP. No two web servers, even ones running the same software, support the same URL structure. You can't throw any random request at any Web server and get a non-404 response. In the same sense no two SOAP servers will be able to do anything with any random message sent to it. So rejecting a message cleanly and informing the client, per the spec, of what went wrong, means you're implementing the protocol, not a subset of the protocol. The pushback on this came from Andrew Layman, Henrik Frystyk Nielsen, and Keith Ballinger of Microsoft; and Sam Ruby of IBM. Thanks.
2. Apps. Remember the search for killer apps we did in April? Let's keep doing that, repeat the loop periodically, systematically, infinitely. One day a month, let's do a roundup of application development in XML-RPC and SOAP. Apps are the reason we're doing the interop work. I'm encouraging people who do toolkits to start thinking about apps. But there's an obstacle in the way for some people. See the next priority.
3. Deployment. This is the new one. I added this to the list as a result of a phone conversation with a toolkit developer, Dan Libby (the same guy we worked with on RSS when he was at Netscape, small world). Dan says he would clone xmlStorageSystem if he had a place to deploy it. I offered space in our cloud, but I would really prefer if someone else worked directly with Dan, instead of adding to the workload here. However, if no one else steps up, we'll do it, because without deployment the apps basically don't exist. (If a tree falls in the forest.. etc.)
There's another side to deployment, I'd like to see scripting environments bake SOAP and XML-RPC into their releases. Make it easier for script writers in all environments to have the same ease-of-access that Microsoft's developers are sure to have. The rest of us shouldn't be playing catch-up as Microsoft deploys. The roadmap is clear, and as interop becomes more solid, the reasons to hold back will be fewer. If you manage a scripting environment, please consider making XML-RPC and SOAP part of your standard install. Even better if operating system vendors other than Microsoft add user interfaces for easy scripting both client side and servers. This isn't as far off in the future as one might think.
4. Business development. Why are we doing all this work? Well, I think most of us love the idea of having lots of choice in development environments, and like creating cool software. But there are bills to pay. UserLand wants to license tools and runtimes. Others want consulting gigs. So last week I added this to the list of priorities. If you'd like to add XML-RPC expertise to your team, we know the people who can do that. If you have a job opening, please post it on the DG on XML-RPC.Com, or send me an email. It's still a buyer's market, I hope not for long.
Wired: RIAA Head says Napster is Done.
Back in SOAP interop-land, Jake announced application-level interop in a demo app that all the implementors are working on called Bid-Buy.
Our president: "Yankee Stadium is hallowed grounds, so is the White House." YYSSW.
Today is Dave Jacobs' birthday. Happy birthday Big Dave!
Linus Torvalds has an excellent rebuttal to Craig Mundie's speech. I love the way he comes right back at them, without too much hissing, with power. Yeah Newton was cool. (Maybe not.) I like standing on the shoulders of giants too. Right on Linus.
Tim O'Reilly: "Microsoft's Shared Source Program is a Validation of Open Source Disguised as an Attack"
Doc Searls: Getting past fear and fudding.
Wes: "In most of the software categories I care about, innovation has pretty much stopped."
Survey: "It's time to choose a platform. Craig Mundie explained the philosophy of Microsoft's platform. Linus Torvalds explained the philosophy of Linux. Based only on the philosophy of the platform, and not their technical merits, which would you choose to develop for?"
Now, after you've chosen, think about another possibility -- not making a choice. It's kind of a trick question. Why can't users who prefer the Microsoft OS get the benefit of your software, if you've linked up to their OS in such a way that you're not locked in. On the other hand, a reasonable person who doesn't want to be locked in could predict that Microsoft, based on past behavior, is likely to do something to get you and your users locked in. Microsoft's pitch would be more convincing if they acknowledged this.
Microsoft doesn't need any fancy gimmicks to be locked in, they already are. The vast majority of users use the Microsoft OS. No need for speeches, no need to arouse the open source faithful. The deck is already stacked in their favor. Just do a sexy OS release, with lots of cool features and eye candy, and then do another. Let developers freely roam the space. Keep your application guys out of the way. Kick back and laugh all the way to the bank.
Register: "In his preamble to his Shared Source presentation, Craig tips the hat to His Billness, who seems now to all intents and purposes the inventor of the Internet." BTW, I have two copies of the original Road Ahead.
© Copyright 1997-2005 Dave Winer. The picture at the top of the page may change from time to time. Previous graphics are archived.