Smalltalk Tidbits, Industry Rants

I've had a few people ask me for my subscription list - here it is. If you grab this, you'll want to delete the Smalltalk-CST folder - the urls for those feeds are internal to Cincom, and you won't be able to get to them :)

It's been a great trip, but I'm ready to head home. My wife and daughter miss me, and I miss them. The With Style guys have been great, and I was able to present to some great groups. The ANU presentation this afternoon went very well - and afterwards, a lot of the folks there were brainstorming about how to make use of RSS and blogs for project management. The Linux group had a few different concerns, but we still had a very good session with lots of good questions. I had a great time, and I'm looking forward to coming back!

Dave points to some pretty anti-social behavior on the part of IIS6:

In a comment on a late post yesterday, Scott Fraer points to a Microsoft article that explains: "When you start Internet Information Services (IIS) 6.0 on Microsoft Windows Server 2003, IIS binds to all IP addresses on the server, not just the IP addresses that are assigned to Web sites."

That's just... wrong

Well look at this - Scoble points to the fact that MS is hiring into the IE team, and that there's an IE blog. So is it possible that IE might support CSS properly?

In the Star Trek universe, there's never too many bad ideas:

Rick Berman, executive producer of UPN's Star Trek: Enterprise, revealed to SCI FI Wire several spoilers for the upcoming fourth season, including the possible casting of original Star Trek star William Shatner (Capt. Kirk) in a familiar role.

Ooh, like we haven't seen this idea before. Over-acting is already a problem in Enterprise :) But wait, there's more!

• T'Pol (Jolene Blalock) may find herself getting married in episode three, Blalock said in an interview.
• The long-standing Temporal Cold War arc will be resolved in the first two episodes.

Will this involve removing the stick from T'Pol's butt? And Berman finally noticed that the temporal cold war idea sucked? Here's a thought - send Berman away on a long term research project and find better writers!

I'll be speaking at the ANU tomorrow at 4 (in Canberra). Then sometime around 7, I'll be speaking at the Canberra LUG (Linux Users Group). Should be a good time.

Sean Malloy shows us how to sell Smalltalk to your boss:

I setup a server, and while test clients were connected to it and sending/receiving messages, I started changing the request handler methods which determined the responses to messages. Simply changing some of the outputs while the test app was connecting to the server. At first it wasn't really apparent to him what was going on. Not until I showed the same task being accomplished in C#. Each rebuild we got to watch the test client start throwing exceptions about failed connections, mean while any change to the code required rebuilding the app. So I wasn't able to change the results of requests on the fly. Trust me when I say if I had been trying to show off C# as the better option, it would have been a pretty bad demo. That was about the time he had a "Holy S***" moment. It was pretty cool.

Heh - I've seen that moment too - it's always great when you get the dynamic nature of Smalltalk across in a visible fashion.

Dare Obasanjo's has noticed something pretty odd in the W3C Compatibility mode docs for XPath 2.0:

The list below contains all known areas, within the scope of this specification, where an XPath 2.0 processor running with compatibility mode set to true will produce different results from an XPath 1.0 processor evaluating the same expression, assuming that the expression was valid in XPath 1.0, and that the nodes in the source document have no type annotations other than xdt:untypedAny and xdt:untypedAtomic.

My reaction is pretty much what his was - huh, what? Compatible isn't compatible, or something. Bizarre...

In a customer visit this morning, I ran across two things that seem to come up a lot:

• Memory tuning (Knowing when/how to muck with ObjectMemory space sizes, and wen/how to tune MemoryPolicy)
• The existence of the double buffering window manager policy in VW 7.x.

The latter is often a simple solution to flicker problems in an application; the former involves appropriate tuning to an application's memory profile so that the GC sub-system isn't doing unnecessary work. In this area, I'd strongly suggest having a look at the appropriate pages of the VW Application Developer's Guide - it's well documented.

Tim Bray wants better we authoring tools:

The state of Web authoring tools is kind of like the state of what we used to call 1CWord Processing 1D twenty years ago when I was getting into this business. If everyone 19s going to write for the Web (and it looks a lot of people are going to) we need the Web equivalents of Word Perfect and Wordstar and Xywrite and Microsoft Word, and we need them right now.

Well heck, I saw those last night at the Sydney Smalltalk Users Group meeting. If you want WYWISWG XML authoring tools for the web, look no futher than Software With Style.

We have some customer visits in and around Sydney today, and then it's off to Canberra tomorrow. Should be an interesting day; then Friday I'm off for home again. It's been a good trip so far.

We (myself and the WithStyle guys - Michael Lucas-Smith and Rowan Bunning) spoke to the Sydney Smalltalk and Ruby user groups this evening. I gave a brief update on the products, and then spoke about BottomFeeder. That was well received - lots of good questions and feedback. Again though - the penetration of syndication is less than everyone seems to think.

Next, Michael and Rowan gave a talk on WithStyle. Rowan gave some background on the work they are doing rich interactivity of desktop apps in the model of web apps. Lots of cool stuff with Rich UI interaction va the use of the With Style engine. Michael is going to do a demo - I'd suggest a quick visit to their website to join their developer program.

Good question from Bruce Badger - some would say that JavaScript "solves" this problem - what say you? Rowan's answer is that JavaScript and the browsers really haven't lived up to that promise. I'd add that lots of people - and firewalls - turn it off. Back to Michael - "We are going to break out of the browser" - we need rich internet applications (ed: like, say, BottomFeeder). They plan to support:

• CSS with XML
• Pollock integratin (with CSS styling!)
• XForms
• Advanced Widgets
• Skinning with CSS

Why CSS - have a look at CSS Zen garden. Now he's off to do a demo. But wait - we're in one - his presentation is in WithStyle. Showed us the example browser, and now he's showing the WYSIWYG XML editor. This is very cool - you can edit the content w/o having to muck around with the XML itself - and it's smart enough to have an open back end (i.e., you can save to arbitrary CMS systems).

Lots of cool things integrated - CSS2 and CSS3 (almost there). ECMAScript integration with Smee from David Pennel. They've got an HTML Window (which I have to integrate into Bf...). Now he's going to show us how to build a presentation app. It's a fairly simple matter of a few methods, or a few files. This demonstrates the integration of XML Events and of Smalltalk scripting - it's a very cool integration.

They are now doing nightly builds and a developer program - you can sign up, get on the mailing list and grab the builds.

Questions: How do you deal with the onset of Avalon and its (someday) ubiquity? Answer - trying to be a rich client solution, not a web browser specific answer.

Thanks to Denis Johnson, I have some pictures of the daytrip we took to Watson bay on Saturday. Here's a shot of Circular Quay, where we we departed on the ferry for Watson Bay:

Here's a shot of the Sydney Harbor Bridge from the ferry. I didn't get a chance to try the bridge walk this trip, which is something I'd like to try out.

Finally, proof that I was in Watson's Bay - a shot Denis took from there. If you look at the Sydney skyline in the background, you can just make out the Sydney Tower. Note the classic VisualWave sweatshirt!

This post in the N=1 blog makes an excellent point about internationalized software:

Now the polite version had the following message when the person did not get the correct answer: "We are sorry that the clues were not very helpful for you. Please try the next game." I can see why some people might think they're being manipulated with such a message. In English, it almost sounds condescending - but of course this was in Chinese, so probably the impact of the message was different.

Which is an important point in itself: there are differences between various cultures, so don't expect that something that works for China would automatically work for North America, and vice-versa.

It's about more that just making message catalogs for different languages - you really have to have someone from the target region helping out. There's another way to deal with this, and I've seen it done with the picture management software that came with our Sony camera. Instead of translating messages, this application had none. Instead, it sports a large number of totally inscrutable icons on buttons. That's definitely not the way to go...

Ed Foster reports a user's experience with Cingular mobile service:

"I have had Verizon for ten years, but I needed phones that would roam in Europe, and decided to go to Cingular," the reader wrote. "After telling the folks at Cingular that I needed the phones for trips being made in the next two months, and being told 'no problem,' I shelled out some $600 for four high-end quad mode phones from Motorola. I confirmed with Cingular and Best Buy (where I bought the phones) before signing that the phones would work when I showed up in Europe, no additional activation charge or notification was required."

The reader then gave one of the phones to his teenage daughter to take to Sweden. "The phone didn't work there," the reader wrote. "I called Cingular to be told that they won't activate the phones until I have established a one year history with them. (I am 50 years old and I have excellent credit.) I told them when I bought the phones I was told that my credit history made this unnecessary. Nope (not sorry, 'nope'), not true. No exceptions. Don't even ask."

I almost switched to Cingular for this trip to Australia (and a trip I'll be making to Europe later this year). It was laziness that stopped me, but it seems just as well - a one year waiting period???. That's just a sick joke. This is exactly the sort of word of mouth story that loses more customers than a company ever realizes...

I'm back from my trip - we had a good meeting with the local Smalltalk users group. Good conversation, lots of good questions about both the product and about my blogging talk. It was again made clear to me that blogging and aggregators are still very much in the early adopter mode. Tomorrow we have another full day of meetings, including the Sydney Smalltalk users group in the evening - meeting up with the With Style guys before that. Should be a good day.

Tim Bray wonders why more people haven't commented on this piece by Jonathan Schwartz, where he states that the commoditization of software is good for Sun. There's a reason that piece didn't get much commentary - watching a clueless Sun Exec whistle past the graveyard just isn't that interesting...

I spoke to the Smalltalk User Group here in Melbourne today - I gave this talk on developments in Cincom Smalltalk, and this talk on Blogs and RSS.

Danny Ayers is talking about getting a unified whole out of RSS/RDF, RSS 0.x/2.x, and Atom. To my mind, he's looking at this at the wrong level - the XML level:

I was just wondering how best to munge data from RSS x.x, RSS 1.0 and Atom. Sure, transforming everything into RSS 1.0 + the odd new term might work, but seems a little messy, strained. There are (potential) equivalences that could be specified properly between DC, RSS 1.0 and Atom/OWL), but that still seems a little limited. That's ignoring the RSS x.x stuff. But it just occurred to me that if a fresh ontology was created, it would be possible to have things like:

rss:title rdfs:subPropertyOf gosl:title
rss2:title rdfs:subPropertyOf gosl:title
atom:title rdfs:subPropertyOf gosl:title
dc:title rdfs:subPropertyOf gosl:title

This is the wrong place to work for that. What you want is to get a set of domain objects - you can work with those in a unified fashion. At the XML level, it's going to be a complete nightmare...

I'll be off for a day trip to Melbourne tomorrow - I'll be speaking to the Melbourne Smalltalk Users Group and doing other visits in the city - I have no idea whether I'll have network access during the day. This will be the first time that I'll have been out of Sydney - I didn't leave the city at all my last trip (in 2000).

Today's weather wasn't quite so good, but I headed downtown anyway. I got to Darling Harbor about 10, and walked down to the Maritime Museum. That was a long, cold walk - it was on the opposite side of the harbor from where I started, and the wind was really whipping around. The museum was quite nice - there's nice US/Australia friendship display at the front - it was put up back in the late 80's for the bicentennial of the founding of the first British colony here. The museum has a lot of other interesting displays as well - lots of information on the harbor, on the history of the 19th century whaling industry, and on the development of the harbor area through time. I spent a couple of enjoyable hours there before going back out to brave the elements.

It was a cold walk in search of lunch - I wasn't about to pick one of the carry out places that expect you to eat at an outside table! I stumbled on a place called "The Meat and Wine Co." - and it was a nice place, with an interesting menu. I selected a nice grilled chicken dish and a glass of wine - they warned me about the hot sauce before I tried it, and they weren't kidding! It was a good, filling lunch. They had an interesting frozen berry mousse concoction for dessert - it was quite good, although I think my wife would like it better than I did - she loves tart desserts.

After lunch I decided to duck into the IMAX theater and catch "Ghosts of the Abyss" - a nice little gem of a movie that documents Cameron's filming of the Titanic. It was something else in IMAX 3D - well worth seeing. The footage of the 1st class dining room windows alone was worth the price of admission - highly recommended.

After the movie, I had to figure out what to do with the rest of the afternoon. I headed over to the Sydney Tower, but the ticket agent convinced me that it was better to wait until dark - the cloud cover would make for a dull view during the day. She whisked out a map and gave me some quick advice. Off I went to Hyde Park, a nice little park in downtown Sydney. At one end is the Anzac Memorial, built after the first world war. The entrances were blocked, with no reason given. It's a very somber building - reminded me of a lot of the Civil War memorials I've seen in the northeast and midwest US. I walked back up the park towards the fountain - the trees on either side of the path offer a very "Cathedral" type framing of the park - very nice effect. Midway down was a spire dedicated in 1857 by the (then) mayor of Sydney, but there was no further explanation. I'm kind of curious as to what it was built to commemorate!

I kep going down the street, passing the Convicts Barracks Museum. I was tempted, but by now it was nearly 4:00 pm, and I wanted to get into the Sydney Museum. It's small (which is good, I only left myself 45 minutes to see it!) - but quite interesting. It has a nice section on the history of the city, including a mockup of the way the city looked in the 1840's. There was a section on the (apparently famous) Red Cedars of Australia, and a lot on the early history of the city. It didn't take long to go through, but it was worth seeing.

I headed out around 5 - couldn't cut back through the Botanical Gardens (they were closing at dark). So I headed back to the Tower, since the sun had set and it was getting dark. Sydney Tower is a lot like the Tower in Toronto, or the one in Tokyo - offers quite a spectacular view of the city all the same. I bought a couple of things to take home, and headed back to the train station. A few minutes on city rail, and I was back at the hotel in Chatswood. It was a nice day (if cold from the wind). I probably don't have any more tourist time this trip - next week is going to be busy with talks and customer visits. I should definitely come back here when it's summer locally.

Blaine Buxton finds Paul Graham off on a good rant about languages:

"I suspect few housing projects in the US were designed by architects who expected to live in them. You see the same thing in programming languages. C, Lisp, and Smalltalk were created for their own designers to use. Cobol, Ada, and Java were created for other people to use.

If you think you're designing something for idiots, odds are you're not designing something good, even for idiots."

Heh - a bit strong - but he's got a point.

Had a nice trip around Sydney today - went down to the harbor and met some friends - took a ferry across to Watson's Bay. We had lunch at Doyle's, which overlooks the water. That was quite nice, but the clouds started to role in and it got cold as we sat there. It wasn't bad for walking though, so we went up the hill to Sydney Harbour National Park, which is a nice place with a great view of the harbor outlet - you've got the rocks below, and the Pacific as far as the eye can see. I still don't have a digital camera, so photos will have to wait. If you want an idea of where I went, have a look here - the guy who wrote that has some nice shots of the area. I didn't go to any of the beaches (it's winter here now, and today was no beach day :) ). It was a nice day out, and I had a pleasant drive back with the local Cincom rep and his wife - who took me on a guided tour of various parts of Sydney. A bit chilly by the end of the day, but it was all very nice. Now I just have to find dinner...

I got this from one of our customers in the education field - it's an interesting take:

I am an associate professor of computer science at Bucknell University in Lewisburg, PA. I am writing because I am interested in getting some "real world" experience using Smalltalk.

I came to Bucknell and started teaching software engineering in 1988. For years I taught my course using C++ and I was always frustrated by my inability to teach software design effectively. C++ was too much of a barrier. When Java came on the scene, I gave it a try. I found it better than C++, but not nearly good enough. Finally, I came across Smalltalk.

I started using Squeak in 1999 and I had an epiphany! I started teaching my software engineering course with Smalltalk in 2000, and I was amazed at how much the students responded to it. I needed to spend a week (or less) at the beginning of the semester explaining the language, and there were no more language issues for the remainder of the semester. Amazing!

The last time I taught my course I used VisualWorks. The students still appreciated Smalltalk, but I realized that I could be even more effective if I had experience using Smalltalk in an industrial setting. I think it would be even better if I had an opportunity to work with a group such as yours which is developing applications in Smalltalk. This would allow me to bring work experiences into the classroom.

That made my day!

Tom Sullivan of InfoWorld asks a good question - when we say that "80% of IT projects fail", we never define failure - it's common to toss around numbers like 60%, or 70%, or 80% - heck, I do it all the time - but what falls under that classification? It'll be interesting to see what he reports.

John Urberg comes to an interesting conclusion on how tools and "culture" affect software development in a given language:

So are there any tools that make it easy to build business objects? How about Smalltalk? Nearly every Smalltalk comes with a real MVC implementation that allows you to bind object properties to controls. Distribution of objects is built into many Smalltalks. You can use the Gemstone database on the backend and all the persistence is taken care of for you. And all this was available a decade ago.

So what's the lesson? Tools have certain ways they want you to use them. Tools like PowerBuilder and .Net want you to pass around SQL result sets. Tools like Java make you write everything from scratch. Tools like Smalltalk want you to make objects. So folks using .Net are going to avoid fighting the tool and assume OO is not a good thing for the business layer. Folks using Java are going to write lots and lots of libraries. And Smalltalk folks will shake their heads and wonder when people will finally get a clue.

Read the whole thing for the introduction to that concludion...

I was pointed to this interview with Tim Bray (of Sun). His comments on RSS are interesting, but I laughed out loud reading this:

Q: You have said that the claims of pundits that .NET is a threat to Java technology's future are silly. And that .NET fails to hit the 80/20 point where you do 20% of the work and see 80% of the benefits.

A: Java 1.0, when it first came out, was very lean and mean. And that was excellent, because people could learn it and become proficient in it quickly. Since then, the Java language has grown into its current, sophisticated, expansive shape. It's going to be tougher for .NET to replicate that kind of successful growth.

Many of the design decisions about .NET were made before it came out. But with Java 1.0, the community collectively was able to build all the extra layers that make up what we have now.

I guess Tim doesn't remember that all the new things in Java 1.5 (or is it Java 5 now?) were either copied from .NET (auto-boxing) or rushed through after noticing that they were part of .NET (generics). He's had the Kool-Aid, just like Scoble over at MS....

Alan Knight is going to be on the OOPSLA panel for The Great J2EE vs. Microsoft .NET Shootout. Others on the panel: Anders Hjelsberg, Don Box, John Crupi, Rob High. I won't be at OOPSLA this year, but that should be a very interesting panel discussion. If you have suggestions for Alan, he's open to feedback - check his blog (link above) for an email link.

I've just had to add a line to my rejects file (which is used to filter referers on the CST blogs) - apparently, some web monkey at the iaea thought that referer spamming was a good thing. Yeah, there's a way to make your organization look good. Bah.

Chris Double points to a source claiming that the JVM ought to be able to support continuations. That's cool, but - based on Sun's concepts of OO, I would guess that they'll invent a completely silly new language construct to support it (assuming they do support it, that is). Why have a simple, extensible language when you can lard on complexity?

Last night I spoke to the Australian Computer Society's OOSig and the Open Source Sig - the topic was Blogs, Syndication, and BottomFeeder. It went pretty well - there was a higher than average turnout, and we had a good discussion on the topics. One thing hit home to me again when I gave this talk - Syndication (RSS/Atom) and News Aggregators are still very much a niche usage. I started out by asking how many people knew about and read blogs - maybe 5 people raised their hands (out of 20+ people). It shrank to 2 people by the time I got to "how many of you know what a news aggregator is?". As cool a topic as many of think this is, there's a ton of end user education left to do. Oh, one last thing - I said I'd get the presentation posted - here are the slides

I notice that Giulio Piancastelli was trying to learn Smalltalk, but gave up - he tried to do this in a workspace:

foo: n
  |s|
  s := n.
  ^[:i | s := s+i.]

He ran into trouble getting a "Nothing more expected" error from the system (Squeak, in this case). Now, for those of us who are Smalltalkers, this is clear - you don't write methods in a workspace, you write them in a browser, in a class. Here, he's trying to write a method in a workspace. Like I said, clear to us Smalltalkers - obviously not clear to a neophyte.

This is an expectations issue. People expect to write code and then compile it "en masse" - which just isn't what you do in Smalltalk. You can write the kind of code he wanted to write in a workspace, but not as a method - a method is typically attached to a class. You can script in a workspace, but you just omit the method entry point. That's what threw Giulio; he didn't realize that he was supposed to create a class, add a method, and then write some script to drive it in a workspace. Probably a better book would have helped - certainly for VisualWorks, a walk through the examples workspace would have helped (as well as a look at the tutorials).

However, that's obviously not enough - VW has tutorials and the examples workspace - what should we be doing that we aren't? I'd like to be able to help people like Giulio not get discouraged.

Gary Short points out how - even though it has libraries for lots and lots of stuff - .NET hinders productivity in critical ways. This isn't unique to .NET - Java has exactly the same problem - sealed classes. Here's a great example:

Here at Scotland On Line a colleague is working on an ASP.Net project. There's a requirement to add a number of working days to a DateTime object. A quick look at the DataTime object shows us that there's no such method. Now if we were using Smalltalk we'd be able to inherit from the DateTime object and add our AddWorkingDays method, but 'cos we're using C# and the library designers have sealed the class, we can't. Grrr!

So now we have to create a wrapper class, something like this.....

namespace SOL.UtilityClasses
{   public Class DateTime
   { 
      private System.DateTime _dt;

      public DateTime(System.DateTime theDate)
         {
            _dt = theDate;
         }
      
      public SOL.UtilityClasses.DateTime AddDays(double NumberOfDays)
      {
         _dt.AddDays(NumberOfDays);
          return this;
      }

      // Implementation of all the other bahaviour we need.

       //Implementation of the new method we need
      public SOL.UtilityClasses.DateTime AddWorkingDays(int NumberOfDaysToAdd)
         {
            //...
         }      
   }
}

...and recreate all the funtionality we need, then we have to add the extra functionality. What a waste of time!!! Why are these classes sealed, come on Microsoft you can trust us, we know what we're doing, you don't have to seal every class in the library!!

This is my answer to those who tell me "but in niche systems like Smalltalk, you have to write your own libraries for things the big vendors provide" - yeah, but Smalltalk doesn't stand in your way and block forward progress. In the case above, a Smalltalker would have added a method to class DateTime and moved on.

The techno.blog quotes a great quote from Alan Kay at his 2003 ETech presentation:

Most software is made by programmers creating text files that are fed to a compiler and loader which makes a runable program. One can tell how early- or late-bound a system is by looking at the things that can be changed while the program is running. For example, it is generally true of C based systems that most changes in the program have to be done by going back to the text files, changing them, and recompiling and loading them. This is even true of Java. Organizationally, this often leads to at most one or two system builds a day before most bugs can be found.

A late-bound system like LISP or Smalltalk can change pretty much everything while it is running -- in fact, both these systems are so good at this that their development systems are written in themselves and are active during runtime. For example, a program change in Smalltalk takes less than a second to take effect, thus many more bugs can be run down and fixed.

But late-bound has some deeper and more profound properties that include abilities to actually change the both the structure and metastructure of the language itself. Thus an important new idea can be assimilated into the constantly evolving process that is the system.

Another aspect of late-binding is the ability to change one's mind about already instantiated structures that are already doing work. These can be changed automatically on the fly without harming the work they are already doing.

Etc. Wheels within wheels.

These ideas are not new, but they are quite foreign and out of the scope of the way most programming is done today. -- Alan Kay.

To summarize - you want a system with no downtime? Use Smalltalk. You want a system that has to be taken down and up every time you make even a small change? Go ahead, use one of the "analyst approved" mainstream systems. I'll be over here, being productive...

Bf has some problems with Asian fonts - in the html display pane. The good news is, assuming that the fonts are installed, the base application is getting the correct fonts (with the fonts installed, I titles are showing up properly). What isn't happening is proper display in the HTML component. I've sent an email off to Holger, the author of Twoflower - hopefully, I'll have a resolution shortly