Some friends and I have been running a study group here on the Island, and we've been looking at the GoF Design Patterns book. Last week we covered the Visitor pattern, and I pointed out that the pattern largely goes away in languages with multimethod dispatch. So I read Surana's post about C# MultiMethods with great interest. His post explains how multimethods work, but not how you would implement them with Type.InvokeMember. I looked at the MSDN pages for InvokeMember, and saw a hairily complex method. Googling didn't turn up anything obvious either. So if there's anyone out there that's used InvokeMember to do multimethod dispatch, and is willing to show their code, I'd be interested in seeing such a thing.

[via Don Box] Dave Stutz has published a new essay titled The Failure of Shrinkwrap Software.

If you've heard Tim O'Reilly speak recently, you won't be surprised by any of the ideas here. I think that the most important thing here is that Stutz worked at Microsoft for long time, has thought about the business that he was in, and come to this conclusion. While the content isn't surprising, he's made a good connection with issues that are relevant to recent security/quality/liability related events.

I'm disappointed that I won't be able to attend BloggerCon this year (yes, I was invited). Our middle daughter, Michaela is turning 3 on Sunday, and there's just no contest about which is more important in the long term. Especially since I missed her birthday last year. So I'll miss meeting fun and interesting people with though provoking ideas and opinions, but I'll be delighted to be celebrating with my little girl.

Ok folks, If you are interested in going to ApacheCon, tomorrow, 9/30 is the last day to get the Hyper-early bird rate of $499. After tomorrow it goes up to $599. So if you are interested in going, click below.

A. San Juan educated me on the meaning of the word monoculture. I didn't occur to me to look it up (and with m-w.com there's no excuse not to, really), so I didn't even think of the relation to agriculture. I was thinking of it more as "single culture", or even more specifically "single worldview". What I'm interested in is the future, and I believe that the future of computing is going to require more new ideas. So a single world view is too limiting. We need the debate on Java/C# vs Smalltalk vs Lisp vs OCaml vs Python vs Ruby vs ??

I've just finished reading The 5 Patterns of Extraordinary Careers by James Citrin and Richard Smith. John Porcaro turned a section of the blogosphere on to this book a while back, so a belated thanks to John for pointing out an interesting book.

As a technical non-executive, it was hard for me to connect with some of what they were saying, but I did find some general principles that was thought provoking for me: the Permission Paradox and the 20/80 Principle of Performance were especially useful.

One essay that is mentioned in the book is Malcolm Gladwell's The Talent Myth which talks about the ways that various companies have failed in spite of or even because of an overabundance of talented people. It's an interesting version of the talent versus process debate. What Citrin and Smith say in response to Gladwell is that those companies may have had talent, but they didn't know how to manage it.

Simon Phipps has a good post with some other postings related to monocultures.

Today's computing environment is far more monocultural than is being discussed. It's not only the monoculture of Microsoft operating systems. It's a monoculture of statically typed languages derived from a C base. It's an operating systems monoculture where most of the ideas are derived or recycled from Unix. It's a hardware monoculture that results in hardware that is optimized to run C programs.

When I got started in computing, there was much more diversity of hardware architecture, operating systems, and programming languages. I frequently find myself longing for those days, and part of what you see in my blog is hearkening back into the history of computing to remind you that much of what's new has already been old, and that some of what is yet to be new, may also already have been old. That's not to say that we haven't made any progress, but maybe it's just a longwinded way of saying that those who ignore history are doomed to reinvent it badly.

Chris Csziksentmihalyi appears on the critical technical practices links page that Simon referenced (unfortunately, I couldn't get to his RPI pages, and his Media Lab pages are a maze that didn't yield the information I was looking for). Chris' work includes bulding technical artifacts that could have existed in an alternate history of technology.

If we took the manufacturing techniques that are used to produce today's high end PC's and workstations, and tried to go back in time to those days of diversity, what would we end up building? Given that we're talking about multiple processor cores on a chip, and hundreds of gigabytes of storage, and incredible rendering performance, would the assumptions that led to C/UNIX and their descendents still hold? It's too bad is commercially infeasible to find out.

Here's some good news: According to Think Secret, Apple's US laptop market share grew 2% in Q2, leaving their total laptop market share at 7%, and moving them to fifth place overall.

Maybe a 15" AlBook isn't so crazy after all...

Last night I was reminiscing about my days at MIT's Project Athena, because I have a T-shirt from there as well. This morning I woke up to news of Dan Geer's departure from @stake. Dan managed (among other things) the systems programming group at Athena, which was mostly students -- some of those students have gone on to bigger and better things -- Ted Tso and John Kohl are amongst the most well know. And they had a cool name, they were called the "watchmakers". I was part of the user consulting group, that was people who were helping people out in the Athena public clusters. We were frequently in meetings that included Dan and parts of his staff, and I remember always being impressed and a little intimidated by Dan. He is definitely one of the good guys.

So I was sad when I read about what happened at @stake. It's a loss for Dan, but he's so good that he won't have a problem finding work. It's a loss for @stake, because no matter what actually happened, people are going to conclude that they censored Dan - and that makes them seem untrustworthy. And it's a loss for Microsoft, for pretty much the same reasons -- no matter what happened, people are going to assume that they leaned on @stake to get rid of Dan. This also looks untrustworthy, and people are looking for excuses to declare Microsoft untrustworthy. All of these losses add up to a loss for you and me, because they'll get in the way of improving security in Microsoft systems, which is a benefit to everyone.

Julie blogged about a conversation that she got into because of one of the old Newton T-Shirts that she has.

Today Chris Double pointed to a post by Mike Lockwood of the Apple Dylan team, which talked about the Dylan shutdown and the T-shirts that they had printed. Well, not only did they print the T-shirts for themselves, they let some of us outsiders, order them (at our expense, not Apple's). So I actually have one of those Dylan T-shirts. The memories and frustrations that one brings...

Update:

Julie just stepped into remind me that I actually have two Dylan T-Shirts, a "regular" Dylan T-Shirt, and a "Cancel" Dylan T-Shirt. And since I eventually did work for Apple, I have a "The Journey Begins" T-Shirt, which is issued at new employee orientation. Actually, I have an entirely too large collection of T-shirts that reflects my employment history...

This probably old hat for lots of you, but I've been using a combination of Eclipse and SecureCRT to access CVS repositories over SSH (at least on Windows). JCraft's CVS-SSH2 plugin for Eclipse makes that unnecessary. It can to SSH public-key authentication, and works great.

[via David Czarnecki's Blog] Mark has written an XPath plugin for blojsom. This is a response to Kimbro Staken's new Syncato blogging tool.

I think that this is interesting because it explicitly decouples they XPath query "API" to the weblog from the underlying storage mechanism. In Syncato, the XPath query API falls out because the underlying storage is XML oriented and provides XPath support (via some code that Kimbro wrote). The thing that seems to have captured people's imagination is the ability to use XPath (some other semistructured query language) to get information out of weblogs at a sub entry granularity. For ease of implementation, Syncato can't be beat (build and install issues aside). But Mark's approach shows that it's possible to get an XPath API to data that isn't stored natively as XML. Will it be more work? Yes. Will someone have to document a "schema" for that data? Probably. The one drawback is that the data in the weblog needs to have enough annotation so that the XPath queries can produce something meaningful. And that may be where things start to fall apart. In pyblosxom, we have entryparsers and preformatters that let you enter your blog posts in a variety of formats. Not all of these formats will have the annotations needed for an XPath engine to find meaningful information. That takes us back to the annotation conventions that Jon Udell was writing about a few weeks ago...

Ben is writing about standards today, but I think that his closing paragraphs generalize to more than just standards setting:

In vibrant markets and political spheres we have means of goverances that can correct misleading or abusive uses of authority. In failed markets or goverments these means may take long painful periods to operate.

One sign of a healthy market or goverment is that those who appear to have authority are much more tentative in using that authority to force the momenteum of the emerging standard. That tentative behavior is the symptom that there are checks and balances in place that temper their power. That tentative behavior signals that they know their status in the system is tenous. That tenous status makes their acts less forceful, less straight forward, more ambigous. That's good, even if it makes them seem slippery, or political.

Or "weak".

Unsurprisingly, there's been some link traffic generated from Greenspun's Java is the SUV... posting. I've see blog postings, and a thread in the SeaJUG mailing list, and the ultimate signal of success, and entry on the Bileblog. I've no problem with dissent, but I'm disappointed by the quality of the critique. Much of the critique is snide "people in acadamia don't know what's going on" or "undergraduates are stupid", or such like. There's a disturbing (to me) inability to admit that maybe there really are some problems with Java, J2EE, JSP, or JDBC. I know when I go to SeaJUG I always hear grousing and grumbling about J2EE this, EJB that. But have someone attack Java as the problem, and suddenly all eyes are blind and all ears are deaf.

What Greenspun said isn't really anything new. Lots of people have been saying the same kinds of things for quite some time. Charles Miller was good enough to provide some code microbenchmarks that illustrate some of the problems with Java. Greenspun's post is the latest commentary on the fact that Java (and C#) have their problems, and that there is still quite a way that we can go when it comes to the foundations that we use to build systems.

There's the easy, simple way to invite people to ApacheCon, and then there's the Ben Hyde way. Only Ben can turn a conference announcement into a learning experience. So on top of what Ben had to say, all I can say is, you better come, because maybe Ben will be there and you'll get some facetime with him. I had dinner with him last year, but didn't really know him, so I missed out. You had better believe I'll find interesting stuff to talk about with him if we end up at the same table again this year.

Recently, Jonathan Schwartz has been saying lots of things that are irritatating open source folks. I'm surprised that no one has picked up on this one. Schwartz discovered that you need MS Word to submit a patent application. Of course, this is a problem for him because Sun doesn't use Microsoft Word. What Schwartz wants us to do is use StarOffice because the file format is an open standard. Of course, StarOffice only costs $79, but that's fine because we're not paying Microsoft for the privilege of submitting a patent application, we're paying Sun, and that makes all the difference.

Open source comes into it because he claims that open source people think that open source will magically fix things like this. Not. Open source doesn't influence which file formats are used by the patent office. And much as Schwartz claims to the contrary, having an open standards file format won't fix the problem either. If the file format were StarOffice, we'd have just as much trouble or more, because most of the rest of the world doesn't use StarOffice and can't write StarOffice files, because the rest of the world is using MS Office, which is closed source and can't be adapted to write StarOffice files. Open standards only matter if they are adopted.

Let's look at more of his argument:

At least for my customers, purchasing decisions are never made on the basis of the licensing convention used to build the product-customers care far more about product quality, price/performance when compared to the competition, and (especially of late) their security attributes. None of those issues have anything to do with open-source any more than whether Sun employs left-handed engineers to build the products, or marketing folks who work weekends (we do, both).

Excuse me, but open standards don't influence product quality, price/performance, and security. If you take the narrow view that open source is about licensing conventions, then I suppose you might come to some of Schwartz's conclusions. But if you view open source as a kind of software development process, they you might conclude that there is a relationship between open source and product quality, price/performance, and security.

Open source ,ay not be relevant to the reasons that the PTO chooses a patented file format for submissions. But its very relevant to the production of high quality software.

Otherwise, the only conclusion you can take away from Schwartz's article is that Sun is the industry's single largest contributor to irrelevance.

Early yesterday I finished the pages for my book. Now there's a slight breather while I wait for the editing folks to do their job. That means I'll have some time to start looking at a bunch of things that have been backed up for a bit, as well as taking a bit of a breather.

Artima has the latest interview with Anders Hjelsberg. The interview focuses on the meaning of virtual in C# vs Java. I'm not passionate one way or the other for things to be virtual by default.

The most interesting part is the second page where he talks about contracts. One of the things that I observed (along with others) at Taligent was that there weren't enough ways to make contracts explicit because there were lots of different kinds of contracts. So here he's talking about incoming and outgoing contracts, which you could call client, and overrider contracts. But when you are building frameworks, the number of contracts doubles for some classes, because you have client and overrider contracts for "users" of the framework, and client and overrider contracts for the internals of the framework itself (given sufficiently complicated frameworks, which we had).

Philip Greenspun has had an epiphany: Java is the SUV of programming tools.

He bases this on his experience teaching Software Engineering for Internet Applications at MIT. I'd have to quote almost all of it, so go read it.

It's Friday, which means its time for the columnists at Infworld to post their latest. I'm impressed by how consistently Jon Udell comes up with something interesting to write about. This week it's themes from Michael Lewis' Moneyball.

The dominant theme is about measurement - mostly that we don't know what to measure.

For me though, these were the "money" paragraphs:

A piece of software isn't just a bag of bits; it's the nexus of a community of developers and users. In the case of open source projects - and increasingly of commercial ones, too - these communities live online. Mailing lists, personal e-mail correspondence, RSS feeds, weblogs, and Wikis are all grist for the social network analyst's mill. By mining such data, this new breed of statistician are able to measure the strength of a person's ties to a community and can assess his or her reputation within that community.

Membership in multiple communities is another potentially valuable clue. In software development as in science, breakthroughs often occur when insights flow across disciplinary boundaries. The conductors of these flows are typically generalists who belong to several (or many) communities and who form bridges among them.

We know intuitively that these are good people to have on your team. Although we can't yet quantify the difference they make, the data, for the most part, is already available for inspection. Perhaps even now another Bill James is making discoveries that will revolutionize what we think we know about the price/performance ratio of software teams.

I think that the second paragraph, about multiple communities, should be, and is going to become really important.