Jon Israelson's Blog

Fri, 19 Mar 2004 20:04:12 GMT

SD West 2004, Thursday (evening) Update Tonight's keynote was given by Bob Cringely and was on the politics of standards. He had some interesting observations on how and why standards are created and some of the motivations (and back stabbing) that takes place. In the end, he advocates the EITF model over things like the IEEE model since it tends to progress faster and gets bad ideas out of the system quickly (fast failure). During the Q&A part of the talk, he opened the floor up to any questions, not just related to his talk. Of course, the discussion went immediately to offshoring of tech jobs. He had some interesting perspectives: - Venture capitalists are sitting on a ton of money (tens of billions of $) that could be used to re-start the tech industry, but they are afraid to gamble on anything other than a "sure thing". Problem is that there is no sure thing in the tech world. - Standardization, like ISO9001, made it possible to move manufacturing anywhere in the world and still be confident that the items produced will be of the needed quality. If you have two ISO9001 compliant plants, one in the U.S. and one in China, with the only difference being cost, well, might as well go with China. The problem is that this model doesn't work very well with software development. First, there are little to no software quality standards. Second, manufacturing is an electro-mechanical process; it's easy to pack up machines and move them anywhere. Software, on the other hand, is a cognitive process, so it's more difficult to move development to a different group of people and get the same results. - There are some cultural issues to be resolved. In other cultures, workers are expected to be "yes men". This can result in problems being hidden until late in the development process, affecting cost, delivery time, and quality. - For some things, you've just got to be there in the customer's face. - With having some faceless person do the work, you get no product/idea ownership. If you're some guy being contracted to do development for a company across the world, then you probably won't be interested in going the extra mile to make the result earth-shattering. I mean, hey, it's just another job in a long stream of jobs. Of course, if the drive and initiative to invent new things also moves overseas, well then we're in really big trouble. - He predicts that we will start to see some spectacular software disasters.

Thu, 18 Mar 2004 20:52:34 GMT

SD West 2004 Thursday (morning) Update I've spent the morning attending classes on AspectJ presented by Ron Bodkin, formerly a member of the original aspect implementation team at Xerox PARC. I've been intentionally ignoring the technology of aspects for a while now, waiting for the technology and tools to mature to the point where I felt that they are usable in production applications. After seeing this morning's presentations, I'd say it's time for me to jump on board the aspect bandwagon. The capabilities of AspectJ and the support provided in tools like Eclipse and JBuilder are excellent (the Eclipse Aspect Visualizer tool is really cool). BTW, AspectJ is just the Java implementation of aspects. There are a number of other language implementations available (Ruby, Python), but the maturity of the tools vary. For the SD West presentations, and others, see the New Aspects web site.

Thu, 18 Mar 2004 01:18:58 GMT

Programmer veterans ponder past, future. Santa Clara, Calif. - A panel of programmers famous for applications such as VisiCalc and Excel and the Apple Computer Inc. Macintosh operating system mused on where programming has been and where it is going during a panel session at the Software Development Conference & Expo West 2004 event here on Tuesday. [InfoWorld: Top News]

Wed, 17 Mar 2004 21:06:49 GMT

SD West 2004, Wednesday (morning) Update Went to two sessions this morning: the first was on .Net CLR internals, and the second was on the "Keyhole Problem". The CLR internals class was pretty good as the presenter was familiar with both the Java VM and the CLR, so was able to compare and contrast them. Short story: not that much difference. VMs have been around for a while, so there's nothing radical in there. The Keyhole talk was given by Scott Meyers and was very good (and funny). Basically, The Keyhole Problem is where software developers place "arbitrary restrictions on your ability to see or express something." He's got some good points. Check out the links for more info.

Wed, 17 Mar 2004 02:18:04 GMT

SD West 2004, Tuesday (evening) Update The keynote at lunch was given by Allan Vermeulen, CTO of Amazon.com. I thought it might be just a big Amazon pitch, but it was actually very good. He discussed how Amazon hires and fosters independence and innovation within its developer community. They do a lot of things that make sense, like giving a small group ownership of a business or technical function, then turning them loose for a month to come up with any type of improvements that they think make best sense for the business. This has resulted in a number of high-value functions being added to their site. I had planned on attending a UML design class during the afternoon, but on a whim decided to go to a course on XQuery. End result: XQuery is a very interesting technology and the big-name vendors are lined up behind it, which is going to give it legs. Jason Hunter, a member of the W3C XQuery group and co-creator of JDOM, gave the class and provided a survey of the language, its strengths, and it weaknesses. The language is functional, not written in XML (unlike XSL), supports dynamic and static typing, can query anything that can be made to look like XML (XML, relational DBs, etc.) and feels and looks a lot like ocaml. It's got a few syntax evaluation quirks, and the whole namespace setup that it inherits from XML can be tricky, but is very powerful. Mark Logic and BEA Liquid Data are the commercial versions to check out today. After the spec is completed, IBM, Oracle, and Microsoft will have something to add. There was a panel this evening called Programmers At Work: Interviews with 19 Programmers Who Shaped the Computer Industry. The panelist were: Charles Simonyi, Dan Bricklin, Robert Carr, Jef Raskin, Andy Hertzfeld, Scott Kim, Jaron Lanier, and Susan Lammers. If you are in software development and don't recognize at least half those names, you must be living under a rock. This was one of the best talks that I've ever been to. There were lots of ideas and opinions about where programming has been, what's been done right and wrong, and where things need to go. The general consensus is that the job of programming needs a lot of revolutionary work to take it to the next level. Be sure to check Dan Bricklin's link, there are pictures and his notes about the event.

Tue, 16 Mar 2004 19:18:28 GMT

SD West 2004, Tuesday (morning) Update This morning's session was on Agile Modeling and was presented by Scott Ambler. It was pretty much a whirlwind of information: we reviewed about 25 types of modeling diagrams. The session was half day compressed version of full day compressed version of a 3 day, feature packed class. It didn't go into much detail, but gave a good indication of which types of modeling is actually used in the real world and where UML 2.0 has gone. He also gave good coverage of the Agile Modeling way of doing things. As I'm waiting for the lunch keynote, I'm sitting outside in the 80 degree, cloudless weather. You can't get any better than that!

SD West 2004 - Monday [UPDATED]

Tue, 16 Mar 2004 00:06:13 GMT

SD West 2004, Monday Update The first day of SD West 2004 has gone very well. I've spent the day in a couple of half day tutorials that cover the basics of Web Services (WS), primarily focusing on Java and .Net implementations. The changes to SOAP 1.2 and supporting elements in WSDL have made WS a much more compelling and complete implementation platform. A key thing to note is that SOAP 1.1 was focused on providing RPC methods, but SOAP 1.2 is is more geared toward messaging (RPC method calls vs. messaging documents). The down side to WS is the HUGE number of WS-*** standards that can be required, depending upon what you are trying to do. Standards include: WS-Trust, WS-Privacy, WS-ReliableMessaging, WS-Federation, WS-Authorization, WS-SecureConversation, WS-Security, SAML, WS-SecurityPolicy, and those are just the security-related standards! Yikes! It's just important to remember that you don't have to use everything; it all depends on the context of the system you are creating. This afternoon's session has shown a lot of security-related code examples using C# and .Net. I haven't worked with either of these technologies, but I have to say that the framework provides some very powerful objects for doing WS. Adding SOAP attachements, encryting/decrypting, signing, using certificates, and so on turn into one or two method calls in .Net. Very nice. I need to look into the corresponding Java tools to see how they compare. Also, some standards are dying or being replaced and new ones are being introduced, so it's important to keep an eye on which ones have momentum and get behind them. Steve McConnell, author of "Code Complete", gave today's keynote speech in which he covered the content of his new book, "Code Complete 2nd Ed.", and gave an overview of how the software development industry has progressed in the 10 years since the first edition was published. It basically boiled down to: the more things change, the more they stay the same. Tonight's round-table is on the merging of SQL, XML, Web Services, and Grids.

Mon, 15 Mar 2004 02:32:20 GMT

I flew into San Francisco this afternoon to attend the SD West conference. The weather couldn't be better: 80 degrees and sunny, and it's supposed to stay that way until at least Friday. Now I'm off to find some seafood...

Thu, 11 Mar 2004 16:38:07 GMT

Java as Baby Step. James Robertson thinks that Java is an interruption in the forward progress of software development. Itís nice to see this meme spread a bit; Iíve thought the same thing since 1996 or so. During ë95 and ë96, watching Java start to gain traction, I was amazed by the ignorance and ire surrounding many of what I considered Javaís best features:

the virtual machine
garbage collection
methods "virtual" (in C++ parlance) by default
a singly-rooted class hierarchy

I came to view Java as a baby step that would serve primarily to soften up developersí attitudes toward these things, thus shortening the leap required to adopt even better languages like Lisp, Smalltalk, and their ilk. I donít think I was alone in believing that, but I didnít hear anyone else saying it for a while.

It is nice to see people returning to serious language research again. Efforts like the Feyerabend Project and more practically focused offshoots like OOPSLAís Onward! track and the Post-Java Workshops (as well as increasing grass-roots interest in languages like Ruby, Haskell, Squeak, Oz, and even an ongoing Lisp revival) give me hope that weíll be ready to take a larger step soon.

[Glenn Vanderburg: Blog]

I'm with Glenn and James on this: I think that Java's success was in part due to it's packaging of the features that Glenn listed above in a form that was easy for C developers to swallow. All of those features had been around for years in various other languages, but those languages didn't resonate well with the general IT population for a variety of reasons. Java acted as the sugar coating that made accepting those features fairly easy, but in my opinion Java has not added any significant new concepts to the world of programming languages. It's time for something better.

Thu, 11 Mar 2004 16:18:45 GMT

The great escape. Immediately after 9/11, dozens of Saudi royals and members of the bin Laden family fled the U.S. in a secret airlift authorized by the Bush White House. One passenger was an alleged al-Qaida go-between, who may have known about the terror attacks in advance. Our first excerpt from "House of Bush, House of Saud." [Salon.com]

I first came across this story while reading Michael Moore's Dude Where's My Country. Basically a day or so after 9/11, when all civilian air traffic was shut down (and I was stranded in Houston), the Bush administration allowed a couple of private jets to fly all over the U.S. picking up Saudi royals and bin Landens and then take them overseas where they would be "safe". A little odd considering that most of the 9/11 al-Qaida terrorists were Saudi citizens and a bin Laden holds the top spot within al-Qaida. Seems like you would want to at least talk to those people to see if they had any useful info. Of course, the Bush family has had very close personal and business ties to both the Saudi royal family and the bin Landen family for many years (read Moore's book for the full story). It also raises the question of why this hasn't been covered more by the press?

Thu, 11 Mar 2004 15:58:37 GMT

Job Scheduling in Java. Scheduling recurring execution of a piece of code is a common task for Java developers. The Timer class has its place, but as Dejan Bosanac explains, developers with more sophisticated requirements might want to check out the Quartz API. [O'Reilly Network Articles]