Smalltalk Tidbits, Industry Rants

Sriram Krishnan desperately needs to look at Smalltalk. Here's where he says we need to go:

Today, the Longhorn PDC build uses up close to 500mb of memory. If you had told this to somebody in the 60's or 70's, they would have been stunned. One thing is for sure...the programming language of the future will let us forget about the hardware underneath.

Hmm. We were abstracted away from the hardware a long, long time ago over here. The Java guys are late entrants to that game, and MS, with their .NET environment, is just starting to get in. Trouble is, Java and .NET still have static types, forcing you to consider essentially irrelevant details. Oddly enough, he's talking about functional languages in his post; someone hand him a Lisp environment :)

Think you have a bad project? Not compared to this guy you don't. He's got to migrate a bunch of legacy C code to Cobol....

Now the fun begins - SCO is finally telling us what they say is infringed:

Certain copyrighted application binary interfaces ( 1CABI Code 1D) have been copied verbatim from our copyrighted UNIX code base and contributed to Linux for distribution under the General Public License ( 1CGPL 1D) without proper authorization and without copyright attribution. While some application programming interfaces ( 1CAPI Code 1D) have been made available over the years through POSIX and other open standards, the UNIX ABI Code has only been made available under copyright restrictions.

Now, we get to see what's real and what isn't...

Joi Ito talks about "blogger's block":

I've had blogger's block lately. As more people read my blog, I realize that I am writing for larger and larger audience. Just about every time I post something, I get thoughtful comments and email from a variety of perspectives. I realize that post early/post often is probably the best policy for blogging, but the rigor in which entries are discussed and the increasing percentage of people who I meet who have read my blog cause me to try to blog about things which interesting yet not something where I'm not likely to have to spend a lot of time defending myself. The fact is, I'm becoming more and more conservative about what I blog.

I know my audience is much smaller than Joi's, but I know what he's talking about - sometimes, it seems that there's too much information, and it's very hard to express anything coherent about it. On the other hand, there are days when it all just flows easily. I do think my writing is getting better; blogging has given me the opportunity to write on a regular basis. Like anything else, you simply get better with practice. For that matter, my typing is improving as well. I still don't really touch type, but it's getting closer

On the other hand, I worry a lot less about this:

he problem with many blogs is that the audience includes so many different communities of people that it collapses the facets of one's identity and requires you to choose a rather shallow facet which becomes your public identity. For instance, I know that people in the US State Department, friends from my Chicago DJ days, my employees, my family, thoughtful conservatives from Texas, cypherpunk friends, foreign intelligence officers, Japanese business associates and close friends all read my blog occasionally. In real life, I present a very different facet of my identity to these different communities, but on my blog I have to imagine how all of them will react as a craft these entries.

Ultimately, I write about things that interest me. Sometimes, other people are interested as well, and they comment - either here, on their own blog, or in email. I can't really try to write for anyone else; either what I write is interesting enough to read or it isn't. I can't really make that call. I'm not really worried about a public/private persona thing either - I purposely limit the topics I post on here. I don't engage in conversations about (electoral) politics, for instance. Why? I do this primarily to promote Smalltalk, and to comment on topics of interest (to me) in the IT sector. This is all about focus; if I started spouting my political philosophy (such as it is), I'd likely offend a number of people who would otherwise visit for news on Smalltalk. It's enough that I generate the occasional tempest in a teapot over static/dynamic typing issues.

Writing a blog is just like any other kind of writing - you have to decide what topic or topics you want to cover, and in what depth. Cover too wide a range, and you may not have enough focus to attract readers. Cover too narrow a range and you may only summon a tiny niche. It's not an easy thing, and I can't say that I really planned any of this out at the beginning. I started this blog as an experiment, and I've been very happy with how it's turned out. Ymmv.

More news from Dave Winer on RSS analytics:

The Clarkbot is a "Perl script written by Rick Heller. It searches the Feedster RSS search engine for references to "Wesley Clark" To be picked up by the Clarkbot, a blog must generate an RSS Feed, and that feed must be listed with Feedster."

Ignore the political stuff, and focus on what this means from a marketing standpoint. Here we have a campaign - and it could as easily be a sales campaign as a political one - aggregating commentary on itself. Why is that useful? Well, that should be obvious. In the past, it was hard to tell what people thought of a campaign without expensive help - polltakers (politics) or analysts (industry). Here's a new method - see what people are writing, and focus further efforts based partially on that feedback.

It's not definitive - but it is one more piece of information, and it's a piece you aren't getting now. Search feeds in feedster are one simple way to get this data - using a bot to scour the net for it is another. The future of analytics is unfolding here

Dave Winer points to the RSS Winterfest - it's a conference on RSS/syndication, which looks like it will be conducted over the net. Fascinating - I can attend without ever leaving my office. The speakers look interesting as well.

I saw a question on runtime deployment and parcel loading this morning in my email; I posted on this

Heh - http://www.itworld.com/nl/ebiz_ent/12162003/. I'll be putting on my flame retardant suit over here :)

Sean McGrath quotes Anders Hejlsberg on distributed development and OO:

The problem with that type of programming OO is: it works great in a single process; it works quite well across processes; it works fairly well in a small intranet; but then it completely sucks thereafter.

...Whereas, we know precious little about how to scale CORBA systems in a geo-scalable fashion. We just don't. There's just no knowledge about it, and I've never heard of anyone being particularly successful doing it.

I'll agree that you can't treat distributed objects "just like" any other objects - doing so will cause problems. That doesn't mean that OO is inappropriate though. I've got a number of customers with nice CORBA based systems - Enron, and now the various successor firms, have a very nice distributed system, in both VisualWorks Smalltalk and java. Having received numerous emails from the developers - and seen the system - I know that it can be done. I also know that it's far easier in Smalltalk - these guys tried it both ways.

Anders' problem is that he's committed to doing distributed systems in a statically typed system, and statically typed systems make it very, very hard to do proxies in a reasonable way. Smalltalk just shines there (I'm sure that Lisp, Python, or Ruby would as well, I just haven't tried). Anders simply has a hammer, and he's stumbled across something that clearly isn't a nail. Thus far, it's apparently baffled him

Greg Reinacker points out a simple way to keep up with what people are saying about your product - subscribe to a Feedster search feed on your product! BottomFeeder has supported Feedster for quite some time now - I subscribe to a search feed on Bf so that I can see what is being said about it.

At the moment, this only captures a small part of the commentary on a product (probably more for a news aggregator, but that's a passing thing). Over time, as more and more analysts and industry authors start providing syndicated content, that'll change. This looks like it could be the start of a route around/beyond the (rather expensive) analyst firms - with the advantage of giving you actual feedback from real users who aren't being actively solicited for their opinions...

After the 5th or 6th serving of potato pancakes, just say no. I just need to find a forklift so that I can move myself....

I've been watching the various presidential campaigns uses of technology - and thus far, the most innovation I've seen in the area is from the Dean campaign. Mind you, I'm neither promoting nor denigrating his policies here; I'm simply interested in what's going on technically, and how it's impacting the world of politics. Most candidates now have blogs, but this news is very interesting - the Dean campaign has built a web based aggregator (which, of course, has an RSS feed). This is an interesting development, as far as I'm concerned, and there are messages in it for smaller companies who want to get their message out - with enough effort, you can create your own channel. It's happening in politics; MS is all over it in the business world. I've been promoting Cincom Smalltalk via the community blog site. Marketing departments, batten down the hatches....

I've been working on some of the issues surrounding character encodings in BottomFeeder today - in the 7.2 based version (for anyone using 3.2 or earlier, this doesn't matter yet). If you are using the 7.2 based version (3.3 beta), then do an update, and you should get some improved display of text in the html pane over time. As a side effect, a lot of items that aren't new will come in as new due to the change....

Dewayne Mikkelson and his Radio WebDog, Shadow has some thoughts on change and software development - too long and too good to summarize; go read the whole thing. One thing this did bring to mind though - CMM. Does movement towards the CMM 5 level also push you back towards a waterfall approach? It doesn't have to, but - combined with offshore outsourcing, I'm thinking it probably does.

We threw a gift exchange party today - our friends and all of their kids (all of them 10 or under). It was fun, but also tiring - a house full of kids from pre-toddler to 10 can be a lot of work - some child is always having a problem of some sort or another :) It all went pretty well though - only one coaster bit the dust, and everyone had plenty (more than plenty) to eat. Now I've just got to finish my Christmas shopping....

Keith Ray writes about the costs of debugging:

From Christian Sepulveda - I was working with a team that had been gradually adopting agile practices, but the developers were resistant to practicing Test Driven Development. At one point, I decided to start tracking how much time was spent compiling code, running the application/debugger, re-establishing the context in the application, observing a result and then closing the application. The developer would think about what he observed, make code edits, start the debugger and repeat the cycle. I observed that it took four minutes on average to run the cycle of compile, launch debugger, find context, etc. This would be done an average of eight times per hour; the remaining hour was spent thinking and making edits

4 minutes compiling and debugging * 8 times/hour = 32 minutes/hour. Imaging how much worse this is in C/C environments where just compiling and linking costs 5 minutes or more before getting into the debugger.

That's one way to look at the cost. Absolutely crappy tools that don't support decent development processes is another. Sure, you should write the tests first, and you'll be more productive doing that regardless of language/tools. But - in a decent environment - Smalltalk or Lisp - the debugger is just a browser that steps through the code, not a standalone, time wasting tool. So yeah, if you use second rate tools, the debugger is a time sink. Find better tools, and it isn't. Imagine how much more productive developers would be...

Patrick Logan asks some difficult questions:

But in the 1970s the typical secretary had a simple tool (Emacs) for helping themselves, the same tool most programmers have intimidated each other from using even as an influence. In the 1980s typical non-technical users were building multimedia applications using Hypercard. Emacs and Hypercard together take a miniscule fraction of the installation space and still a small fraction of the intellectual power required for computing with XML, DOMs, XAML, and WS-xxx. Are the secretaries going to be doing this in Info Path?

That's a good question. There's little doubt that the scions of the industry have been making life more, rather than less, complex for us all. Smalltalk, Lisp, and Emacs are loads simpler than what Sun, MS, and IBM have been tossing at us - one has to wonder about that....

My wife bought me a new CD as a present, and it's from an artist that I hadn't even heard of before - Tara MacLean. The album I'm listening to is "Silence", apparently from 1996. It's excellent stuff; very good music to work to. Not that I've tried the site out, but my wife recommends the Tower Music website - she found this album (and a bunch for herself) by listening to samples off the site.

The European Smalltalk User Group is proud to announce that it will organize an academic track for the 12th year of existence of the ESUG Conference with an excellent program committee.

Scope:
The goal of the academic track is to have a forum for publications related to Smalltalk and dynamically-typed languages. We encourage authors to submit excellent quality papers as we plan to produce proceedings. The organizing committee strongly discourages the submission of product presentations and other marketing related material. The academic track is about research!

A non exhaustive list of topics is

• new languages features (mixins, AOP,...)
• meta and reflective programming
• code analysis (refactoring,...)
• process development (Agile processes, Unit testing, ...)
• virtual machines (optimization, new trends, ...)
• frameworks (web, graphical...)
• software evolution (metrics,...)

The best papers will be published in a special issue of the Elsevier international journal Computer Languages

Program Chair:

Dr. Noury Bouraqadi (Ecole des Mines de Douai, France)	email
Prof. Stephane Ducasse (University of Berne)	email
Prof. Roel Wuyts (University of Berne)	email

Program Committee:

• Prof. Andrew Black (Oregon Health and Science University, USA)
• Dr. Gilad Bracha (SUN, USA)
• Dr. Noury Bouraqadi (Ecole des Mines de Douai, France)
• Prof. Serge Demeyer (University of Antwerpen, Belgium)
• Prof. Theo D'Hondt (Universiteit Vrije Brussels, Belgium)
• Prof. Christophe Dony (University of Montpellier, France)
• Prof. Stephane Ducasse (University of Berne, Switzerland)
• Dr. Robert Hirschfeld (Gomoco, Germany)
• Prof. Ralph Johnson (University of Urbana Champain, USA)
• Prof. M. Marchesi (University of
• Prof. Kim Mens (University of Louvain la Neuve, Belgium)
• Dr. Serge Stinckwich (University of Caen, France)
• Prof. Dave Thomas (Bedarra, USA-Canada)
• Prof. Roel Wuyts (University of Brussels, Belgium)

Important Dates:

• Deadline: 21 of May 2004
• Notification of acceptance: 1st of July 2004
• Final version: 21st of July 2004

Format information:

• http://www.esug.org/summerschools/2003_Bled/elsevier.sit
• http://www.esug.org/summerschools/2003_Bled/elsevier.zip
• Preferred format: PDF
• Maximum paper length 15 pages

How to submit a paper:
Send your paper in pdf format to Noury Bouraqadi, Roel Wuyts, and Stephane Ducasse

Via Derek's Rantings and Musings - is there a ghost in Hampton Court palace, or just a costumed worker behaving oddly?

Gordon Weakliem chimes in on the topic of editors and development tools:

I've suspected for a long time that many, if not the majority, of developers at Microsoft, at least on product teams, don't use Visual Studio. I'd bet that most devs at Sun don't use SunOne, or whatever they call it these days, either.

...

What aggravates me about Eclipse is that sometimes, you just need to edit a file; in Eclipse, the file has to be part of a project. I know that this helps avoid problems where you end up with resources that aren't source controlled, but it's still too restrictive for me.

This is starting to look a lot like a cultural issue to me. We Smalltalk developers have been known to say (and I'm sure others find it annoying): "Source code in files. How quaint". There's a reason we say that though - in Smalltalk - unlike most other environments - the development environment is the deployment environment and vice versa. There are no files to aggregate, edit, and compile; there's the image (yes, I know some Smalltalks are different; I'm talking about the more commonly used ones). What that does is create a very different development world view. Additionally, our methods tend to be short - for me, methods longer than 7-10 lines of code start to positively smell. How does that tie in? Well, just how powerful does the editor need to be when you are dealing with 7-10 lines (at most) most of the time? Not very. This is one of the reasons that power tools like Emacs just bore me; The Smalltalk environment is more programmable, and I'm not dealing with large amounts of text anyway.

Bottom line, I think the folks dealing with curly brace languages have a different view of development from Smalltalkers, and part of it involves needing/wanting a more powerful editor.

If you tried to download CST non-commercial this morning, you ran into server errors. I have been trying to diagnose a bug in a new servlet I'm implementing, and I was running the system headful (with a GUI via ssh and X11). As it turns out, there was another bug in code that had been added to the NC app a week ago, but had not reared up yet (only came up at startup). Things are back to normal now; I apologize for the problems.

It's going to be a busy couple of days here - we have all the normal Christmas stuff to do - shopping, decorations, tree (etc.) - plus a gift exchange party with our friends (tomorrow!), and then my parents coming on Monday for a week. Add in a Hanukkah dinner with my in-laws this weekend, my wife and daughter heading to Hanukkah services tonight (is this multi-cultural, or what? :) ) - and it starts to look hectic. Then, my daughter has girl scout events today and tomorrow, and school report due before the holiday break.

I thought the whole purpose of time was to prevent everything from happening at once?

The Pomo Blog reports that the Red Sox used blogging during the (failed) negotiations for A-Rod:

The old-media types are furious - but the Red Sox regularly posted on a fan site/messageboard to discuss the now-dead A-Rod for Manny Rodriguez trade. This culminated in a hilarious showdown on a local sports radio show. One sports writer was complaining how the Red Sox were trying to go "directly to the fans." New Sox pitcher Curt Schilling called in - shut the writer up - and told him the web is the place to go if ballplayers want to get their story out unfiltered.

I wrote about testing awhile back, and that generated a few responses from Ryan Lowe. Interestingly enough, one of the things we were discussing just bit me this morning. A little background first.

The engineering guys have been making the VW installer into a network capable installer - a really nifty thing, and something that will end up being part of our (eventual) support for auto-update. We are planning to deploy the net installer for use in downloading Cincom Smalltalk NC - the idea being that you could either download as you do now, or grab the installer and pull down what you need at install time. Bottom line, engineering whipped that up pretty quickly, and has been waiting for me to have a servlet that would allow for registration from the installer

Now, I already have the servlet used by the registration app. However, that servlet has some inherent coupling to the workflow of the online application, which would differ somewhat from the workflow of the client application - even though they both talk http. So I created a new servlet, and a test version (that stubs out some of the things that require server side access) for local testing. Well, one of those stubbings out - of something I thought "couldn't break" - broke.

So now I'm adding a few new tests to cover the stuff that "couldn't break". The good thing is, I'll have a more stable application at the end of all this.

Joi Ito relates an interesting story about some work he recently did for a government entity in Japan:

I was asked to take a look at the audit and provide a 3rd party opinion. Since I am on the central government panel working on the security of the Basic Residents Registry, my letter has become a bit controversial and apparently my phone is ringing off the hook right now in Tokyo. Lucky for me I'm in the US...

This seems to be an all too common reaction to unpleasant news. An entity (private or public) brings in a third party for an independent review of something. Said third party does the review, and reports some problems. The original entity then has a hissy fit over the report.

Do these people even realize that the public fit over the contents of the report only make them look worse?

There, let me slap the post into the marketing category and maybe this will be clearer. Don Box responds to my post on eating your own dogfood. It's interesting that Don apparently uses VS in private and then publically uses a text editor in a marketing demo. I watched that demo with a bunch of other developers recently. More than a few walked away from it saying "boy, VS must suck if even the MS guys won't use it". This is all about perception, not reality. What you do in a public marketing demo has consequences. Regardless of what message Don wanted to send, another one implicitly went out.

A Buffy/LoTR crossover - and it's a Musical....

Ted Neward reiterates some painful - but frequently ignored - truths about the browser and apps that try to live in it:

At the end of the day, we need to stop pretending that the Web browser is just an extension of the desktop, and recognize it for what it really is: the old mainframe terminal, gussied up by lots of colors, graphics, and better fonts. Treat your web apps as we did terminal-based apps 30 years ago, and you'll find your system behaves a lot better as a result.

The amusing part of this is, we started writing web apps right at the end of the era of migrating from green screen apps. Think about it - green screen apps to Client/Server to green screen with graphics. The sheer size of the money pile spent on all of that is awe inspiring...

ArcterEx comments on my post hitting Scoble for publically using Notepad to code. He gets into a lot about understanding the machine, which really doesn't have a lot to do with what I was on about - so let me reiterate. Scoble is, best I can tell, a marketing guy. What his post did was loudly proclaim that the MS development tools are too hard to use - even if you work for MS, and presumably have direct access to the developers. That's a terrible message to send out to the market. I don't code in vi and file-in all my code - it would be a very negative commentary on Cincom Smalltalk - just like Scoble's post ended up being a very negative commentary on Visual Studio. What this amounts to is an implicit marketing failure - made worse by Scoble clearly not getting it. Make no mistake - anyone reading Scoble's blog the last couple of days got the idea that VS is not a good tool for beginners. Somehow, I seriously doubt that this is an idea that the VS team wanted to have MS marketing pushing out....

On a related note to my last post comes this missive on J2EE from Glenn Vandenburg. Here's the gist of it:

Here's what I think the problem is: we started with a language that is insufficiently reflective (so that it can only be extended in limited ways from inside) and that has too much syntax (making it difficult to adapt to more declarative tasks). So we have containers, code generators, interface generators, EJB compilers, and bytecode enhancers, XML config files running out the ears, and on and on.

It doesn't have to be that way. Try building web apps in Ruby, using Borges. Or Smalltalk, using Seaside2 or Comanche. Or in Common Lisp, using KPAX or WebActions.

Or VisualWorks Smalltalk, using Web Toolkit. Creating SSP pages and servlets in Smalltalk is a model of simplicity. You have to wonder when (or even if) developers will start to realize that there are better ways than the Java and C# paths to heck...

Charles Miller points out one of the rarely talked about complexities of J2EE development - the need to know multiple development languages. It's all just Java? Not exactly, says Charles:

A coworker and I once tried to work out just how many different languages you had to know to understand all of a J2EE project. Between programming languages, markup and templating languages and a thousand different flavours of configuration files (each a language in its own right), the assumed knowledge for a pretty simple project was pretty daunting.

Now to be fair, most application development will involve at least 2 languages - the main one for development, plus C/C++ (for any linking down to platform capabilities you need to do). Also, if you do web development, you can count things like XSLT and CSS as languages to learn in that regard. Still, the point holds - Charles is saying that J2EE development forces an awful lot of complexity at you without regard to the type of application you build. Here's where the problem hits:

You don't see nearly so much of this with other languages. Not just the proliferation of different languages, but also the proliferation of configuration files. Everything that's optional or that needs to be easily changed (scripted) gets dragged out into something that isn't Java code, where in other languages it would much more likely be written in one language, and programatically configured.

A configuration/scripting file makes sense when it's one component and one config file. But when you start gathering components together, you end up with one or more configuration files per component, and things start to get brittle and fiddly: especially if you want to make regular switches from one configuration mode to another (for example testing and production), in a way that crosses a number of different files.

Charles explains this tendency as not being a Java thing per se, but a cultural thing - a legacy carried forward from the C development mindset. That's interesting, because it gets right into something Joel was on about the other day - cultural differences between Unix and Windows development. I think Charles is correct, and there are cultural differences between groups of language developers as well.

Scoble justifies not using Visual Studio while learning C# this way:

I'm learning to program. Not learning to use Visual Studio. I wanna know how to branch my code. How to properly call objects. Send around strings. How to write algorithms.

Why would I want to do that in Visual Studio when I'm a beginner? To tell you the truth, I also am playing with Visual Studio as I work through the books that I'm working with. Why? Cause I wanna see what the experience is like in both places. But, Visual Studio is an awfully confusing place to be for a beginner. Tons of icons. Debuggers. Properties. Options. What again do those have with writing a loop or dealing with strings?

Now that's amusing. VS is too confusing for beginners? If that's the case, then there's a problem - MS spends gobs of money on studying user interactions and user suitability, and they've ended up with the same issue we fret about with VisualWorks - only we have virtually no marketing budget, and no budget for user studies. If the tool is too hard for beginners, then that's something that needs addressing, and fast. We are trying to do that with VisualWorks - I think Scoble should find the relevant people on the VS team and explain why VS is too hard for people like him to use while learning. Trust me, they would appreciate the feedback. The whole point of an IDE is to make cumbersome tasks easier. If it's not doing that, it's a problem.

Since he was pointing to this link from his blog. Heh - when was the last Red Sox World Series victory again?

Interested in the development of withStyle? Then subscribe to this feed

Sometimes, storage matters

I came across two interesting posts over the last two days, both of which the authors eventually thought better of - they deleted them. I'm not going to link to the cached posts; the authors deleted them from their blogs, and it's not the content of the posts I'm really interested in here. What is interesting to me is this:

• For blogs, there's Feedster. Feedster is a lot like Google, but focused on blogs. Like Google, Feedster has a cache
• There are lots and lots of people reading blogs via aggregators. These aggregators tend to cache posts locally.

What does that mean? It means that comments you post - even briefly - will likely get cached all over the place. Even if you later regret what you wrote and delete it, lots of people will still have full access to your full, original comments. Much like email after you hit send, you should operate under the assumption that anything you post is permanent.

Via Dave Winer comes word of a feed tracking New Hampshire Primary Candidate appearances.

Matt Croydon has details on what you can do about it

Richard Monson-Haefel notes the complexity of J2EE development, and says that it needs a simpler programming model:

Over the past four years the various J2EE APIs (EJB, Servlets, JDBC, etc.) have become more and more sophisticated and, unfortunately, more complicated. As a result the learning curve has become ridiculously steep " for every API in J2EE there are dozens of types and hundreds of methods and a bazillion books designed to make them easier to understand. The increase in sophistication is necessary because enterprise software is inherently complicated, but do the APIs need to reflect the complexity of the underlying systems or should they hide the complexity? I argue that the J2EE APIs should reflect complexity as well as hide it by offering both implicit and explicit programming models.

Now, complexity of this sort is hardly limited to Java, even if J2EE is a particularly bad example of it. It's what happens when you allow a bunch of people who don't seem to understand distribution to build a framework around it....

Scoble writes about his first .NET application, written using Notepad. Of course, Don Box led the way by demonstrating .NET and XML based extensibility by using a DOS Window and Emacs. Some of us prefer productivity. When a marketing guy talks about development, and uses Notepad instead of your primary development tools, that doesn't say much in their favor. Doesn't even matter whether VS is good, bad, or indifferent - it's a perception thing. Have influential people visibly not using your development tools, and it makes people wonder.

Scoble links to some criticisms of him (here by me, and here by Dare Obsanjo) - and then to some controversy in Microsoft over Dare Obsanjo's moving of RSS Bandit to SourceForge. It's an interesting thing - the various MS blogs are putting a very human face on MS - and the lack of constraints from MS senior management is helping their image with the part of the technical community that reads blogs.

It was great. I'm looking forward to the DVD for some of the cut scenes (the ones with Saruman especially) - but wow. Great job by Peter Jackson. Run, don't walk, to the nearest theater!

Blogs must be going mainstream - this lawfirm has a blog, and over here, a judge has a blog, complete with an RSS feed. Wow.

I'm off to see the midnight show of ROTK - after I take my daughter to her flute and dance lessons, and have dinner. I'll be tired tomorrow, but it will be well worth it

The Register reports that intel is caving on the "year of itanium":

A few gallons of rancid egg nog were poured this week all over the "Year of Itanium" celebration underway at Intel, as an analyst firm predicted Intel will not only give in and ship a x86-64bit chip but also that the product will be woefully behind in the market.

...

"Our research suggests Itanium is in for a rough ride," Whittington writes in a recent research note. "Intel is now saying it will "go with the market" on 64-bit x86, thus is destined to unveil one when they think the market will ripen, which we judge as mid-2004 for volume delivery in 2005.

Why is intel going to make this move? Well, the story goes on:

Intel is edging close to breaking the 5,000 servers shipped per quarter barrier. With robust sales right around the corner, why would Intel introduce an Itanic competitor now?

Well, it so happens that AMD is already moving more than 10,000 Opteron boxes a quarter. That's more than the total number of Itanium boxes shipped all year. If this x86-64-bit thing is taking off, Intel really would prefer not to be left too far behind. Add to that Sun Microsystems' recent entrance into the Opteron market along with IBM's support, and things start to seem a bit worrying.

Looks to me like intel made a serious slip when they went down the Itanium path

Brad Abrams writes about the learning curve for languages and libraries:

One of the pearls of wisdom I took away from yesterday's class was Anders' assertion that when the C language was developed the learning curve for most developers was 90% in learning the language and 10% in learning some libraries. But today with C#, VB.NET, MC++, etc the learning curve is more like 10% for the language and 90% for the library. (Well, OK, maybe it is 20% for the language in the case of C++ ;-)). The relevant point for my class was that if most of the learning curve is in the library then to make the system easier to learn we have to focus on the library. It is also worthwhile to think about what the implications of this assertion is on the great language debates (C# vs. VB vs C++ vs Eiffel vs Cobol vs Ruby vs Perl vs...)

This is true, as far it goes. The problem with libraries in the Java and .NET worlds is that they are assumed to be complete by the original developers - many of the classes are defined as final, so that you can't even subclass them, much less extend them. There's a flaw in this reasoning though - the flaw is the assumption that the library developers know all the possible use cases for their code, and have considered and allowed for all of them. That's simply not the case - just go look at all the bits that have been spilled in Java due to the various final classes, for instance. This is why - even though most of the learning is at the library level - the language and runtime still matter. In Smalltalk, one can extend or change any class or method you need to change. That's highly useful - because the library designers don't always make it easy to replace a given class via subclassing. Sometimes, you just have to go in and modify the base system (or build scads of wrapper classes). In BottomFeeder, I have had to both subclass and modify the base XML parser - in order to handle errors in ways that the original developer simply did not foresee.

It's a nice theory to say that the library can be fixed, or that an unchangable library is a boon - but that sound you hear is the sound of lost productivity

Dare Obasanjo takes Robert Scoble down a few notches. I had much the same reaction as Dare did when I read Scoble's post, with an additional notion that reminded me of my days as a sales engineer - Robert is starting to sound like a sales guy who thinks he's technical....

Charles Miller has some words of advice on the uses (and over uses) of mock objects and unit tests:

Most of the systems I've seen over the last few years have been written in layers. Down the bottom, you have components that do the important stuff, but don't talk to other components much. The layer above coordinates between the lower-level components, but doesn't do much interesting itself apart from that delegation.

So what happens if you test those higher-level components with mock objects? Well, you sort of end up with tests that do absolutely nothing but test a series of method calls on mock objects. This is, I would suggest, worse than useless. What are you testing, really? You're not putting forward any kind of verifiable hypothesis about what the method you're testing should be producing: you're just writing it twice, once in the test and once in the object being tested.

Which seems to say, only test the stuff that matters - something that harks over to what Ryan Lowe was getting at the other day. The hard part, of course, is that different people will draw these lines differently.

I've been working on something experimental for BottomFeeder. I've got code for eliminating the feed list folder completely - at startup, and feedlists would be folded into the subscribed folder, and from that point forward, any feedlists added would come in as folders and feeds. No more artificial separation between feeds and feedlists, it all just becomes one seamless whole.

Now, I haven't versioned this into the public store yet, nor have I deployed it yet. What I'm wondering is, how useful is this? I rarely use feedlists, so I'm not the best judge of this. Any Bf feedlist users want to chime in?

Ryan Lowe makes some good points about outsourcing, and how to deal with it.

Dare Obasanjo explains - unofficially - some of the rationales for censoring benchmarks:

I'm not sure what the official line is on these contracts but I've come to realize why the practice is popular among software vendors. A lot of the time people who perform benchmarks are familiar with one or two of the products they are testing and know how to tune those for optimal performance but not the others which leads to skewed results. I know that at least on the XML team at Microsoft we don't block people from publishing benchmarks if they come to us, we just ensure that their tests are apples-to-apples comparisons and not unfairly skewed to favor the other product(s) being tested.

Yeah, I understand that theory. I'd rather be open, and meet bad speech with more speech though. Regardless of your reasons, having a license that forbids benchmarks without prior approval is going to come across as hiding something

Larkware News talks about Log parser, a useful sounding tool:

Q. What is Log Parser?
A. It's a command-line tool from Microsoft that lets you run SQL queries against almost any sort of Windows log file, and get the results out to an array of destinations, from SQL tables to CSV files.

If it works as advertised, that sounds like a useful tool to add to the bag.

ComputerWorld has an article on CMM and outsourcing. They have some interesting thoughts on the mismatch between organization types:

"CMM is a great discipline, and it is a great designation to have," says Bart Perkins, a Computerworld columnist and managing partner at Louisville, Ky. based Leverage Partners Inc., which helps CIOs manage IT suppliers. "But the reality is that if an outsourcer is at Level 5 and the client is at Level 1 or 2, the client doesn't have the internal discipline to take advantage of the Level 5 provider's standardized routines."

Defining system or project requirements is a prime example. "With CMM, the entire requirements process is very rigidly defined. A Level 5 requirements document is very detailed and explicit and has metrics associated with it," Perkins explains. "But a company at a CMM Level 0 or 1 could have their requirements on the back of an envelope and no metrics. The Level 1 companies are lucky if they write out two pages."

The upshot, says Perkins, is that touting a CMM Level 5 rating to a Level 1 buyer "comes down to touting a feature that's of little value. It's like a car salesman in Alaska touting a car's great air conditioning. It may be great, but you can't take advantage of it."

Heh. I seem to recall that only a few years ago, the completely specified, over the wall to IT and back again later methodology of development waqs being denigrated. So it's a good thing if we can get software that doesn't meet our actual requirements, so long as it comes in cheaper? Organizations that don't flesh out a detailed requirements document with metrics aren't necessarily broken - it may be that they are being honest with themselves about how much they know ahead of time. What this tells me is that CMM 5 done overseas is just BFUD, only with less communication with actual users. That's good how, exactly?

Ed Foster points out that MS - like many other vendors - is forbidding benchmarks as part of their standard contracts:

Is it possible Microsoft has something to hide about the performance of its server and developer products? It's hard to escape that conclusion when you see how many of its license agreements now contain language forbidding customers of those products from disclosing benchmark results.

The Cincom Smalltalk team is always happy to see customers run and publicize benchmarks, even when they don't look good. We don't pretend that we can run our software under all the types of use cases that our customers do; the only way we can improve is if we know where the problems are.

The MS clauses are worse than most; take a look at this:

Note that Microsoft's censorship clause doesn't just say you can't publish benchmarks " it says you can't disclose them at all, presumably to anyone, without written permission from Redmond. That leads to all sorts of interesting hypothetical questions. Suppose, for example, your boss asks you to run some performance tests on Exchange Server 2003 to see if your company should deploy it. Can you disclose the results to her without bringing Microsoft police knocking at your door?

So what are MS and the other vendors afraid of?

Dave Thomas is using RSS to track his to do list:

My most recent experiment was with the simple ToDo system in Apple's iCal application. I ended up with lots of ToDo's in there, but unless I left the iCal window open all the time, I kept forgetting them. I just didn't look at them often enough.

So, I spent the last 15 minutes writing a quick hack: a Ruby CGI that takes the ICalendar format .ics files that iCal produces, extracts the VTODO items, and reformats them as an RSS feed. I then subscribed to this list inside my aggregator, NetNewsWire. So far, I like the result: my ToDo items are just another list in my aggregator

That's a great idea. Unfortunately, I don't keep my to do's electronically, so I can't just parse them to RSS...

Thornton May at Computerworld has a weird idea:

Herein lies the disconnect. McNealy and his senior team are spooky smart and have thought hard and long about how our industry works. The top of the house at Sun has architected a plausible, affordable and practical alternative path to the Microsoft hegemony.

Unfortunately, this message isn't the one being delivered by the Sun sales force, which is perceived by most IT leaders as being little more than coin-operated box sellers. Sun's sales force is the least influential and respected of all the major vendors.

This is just too amusing. So Sun's sales guys are selling the wrong stuff, and it's because they lack vision? Let me explain something very, very simple about sales reps - they sell based on their compensation plans - i.e., they try to sell what will best reward them. Who sets those rewards? Why, that would be senior management - you know, those guys that Thornton May seems to think are Spooky Smart. So smart, in fact, that they have set up the sales compensation plans so as to prevent the sales of the things they claim to want to sell. Do I know this for a fact? No, I haven't seen the Sun comp plans. However, I have worked with direct sales staff for years, and I know how they operate. If Sun's reps aren't selling something - across the board - then it's an artifact of their compensation plans

I think Thornton May needs to go find a different set of adjectives to describe Sun's management.

Through the magic of ReplayTV, I just finished watching the two day miniseries Battlestar Galactica. I watched, but was never entralled by, the original series (and don't even ask about that last season - bleah). This was much better. Interestingly enough, the effects were not really the best part of the show - the characters actually had some depth. It's clearly trying for a prime time series slot, and I think it would be worth it. Other than SG-1, there's just not that much good sci-fi on tv right now (and no, Enterprise doesn't count as good sci-fi).