Smalltalk Tidbits, Industry Rants

Don Park is more down on Atom than I am:

For example, what is the point of requiring all entries to have a unique identifier when feed clients have no incentives to reject feeds that violate that requirement? Filtering of duplicate entries from multiple sources is a quality of service problem that is best dealt with in a separate opt-in specification. Clients and servers that support the opt-in spec will be more popular if resulting differences in user experience and distribution are significant enough. If not, there was no need for it anyway.

The reality is, aggregator implementors are not going to charge out and ponder the spec in great detail. We are going to do a sanity check on Atom 1.0 feeds as they appear, tweak existing code as necessary, and move along. In BottomFeeder, RSS feeds, RDF feeds, and Atom feeds all end up being pushed into the same object model - from the end user's standpoint, a feed is a feed, and whether it's Atom, RSS, or RDF just doesn't matter.

The net value of the entire Atom exercise? Very close to zero.

As you know if you read Troy's blog, he's off having gastric bypass surgery. I've been getting email updates on his status, and this morning I found out that he had been suffering a small fever since just after the original surgey. Today they went in to check the sutures - they wanted to make sure that there was no infection caused by leakage. The good news is that there's no leakage, and his fever is going down. Here's the update I got:

Troy is out of surgery, and they did not find any suture leaks, or any problems with the surgery. They are not sure what caused the fever, but it seems to be going away this afternoon.

So good news for Troy!

When I announced CST Summer 2005 yesterday, the NC files for ObjectStudio weren't updated on the server - the 7.0 files were still there. That's been updated, so you should now see ObjectStudio 7.0.1 as one of the download options.

May the farce be with you :)

PVRBlog links to a Steven Levy piece on TV watching habits:

The result may be that when all the time-shifting and space-shifting is accounted for, most people will watch the same stuff by the same creators. In fact, even with today's relative abundance, most people stick to only a few channels. According to Nielsen Media Research, households that receive about 60 channels usually watch only 15. Households whose systems can receive 96 channels (around the national average) actually watch... 15. What's more, a recent study conducted at the UPenn Annenberg School for Communications showed that when people were offered more programming choices, they stuck to fewer selections—and, alarmingly, watched fewer news shows.

Well, the thing is that those "15 channels" vary a lot, I think. While each of us may only watch a small set of the available stuff, I'd bet that the small set varies quite a bit over different households (and varies quite a bit over time depending on the household demographics, as well). For instance - we watch a lot less Nickleodeon now that our daughter is 11 than we did when she was 5.

The hard bit of market research here is figuring out what part of the target audience any particular household is in.

Here's a blog that's all about PVR news - the PVR blog. Subscribe to the feed here.

Steve Rubel points to an interesting fake blog occurrence - where whoever is behind the fake blog is spamming negative comments about the company in question with positive spin. Fake blogs are getting more common - take a look at the results of this Google search on the term. Of the supposedly 10 million + blogs out there in the US, I wonder how many are either:

• Dead (i.e., one post and gone)
• Troll blogs of the sort I mentioned above

One things for sure - it'll be really hard for search engines to filter this out - not least because the service being abused most is run by Google...

Update: The faux blog has gone 404. I guess Panera Bread doesn't like being caught out :)

Is this the state of thinking in the Java world?

I spend half my time trying to identify what performance systems are doing by reproducing their behaviour in a performance testbed. It isn't always successful - I'd like to get profile information directly from the production systems, but even low overhead profilers can't get certain types of information in a low enough overhead way (I've tried, believe me I've tried).

One piece of info that is really useful to know is how many of each type of object has been created and has been GC'ed by the JVM. That information isn't available from the JVM, except if you use a profiler (with the honorable exception of JRockit which I think will give you this info in a low overhead way).

But all it takes is two instance variables added to each class object: countCreated and countGCed. And then the JVM just needs to increment each counter for each class when an object is created or GCed. That must surely be a neglible overhead added to the cost of object creation and garbage collection.

This is really the biggest bang for your buck I can think of to add to the JVM for improved monitoring of the system. With that information, so many things become possible, including automatic detection of memory leaks, identification of what may be causing high GC loads, etc.

Gads. Does this guy have any idea how many instances of an object could get created in a normal application? Remove your hands from the code browser, and return your seat to the full and upright position....

My windows inside Eudora were disappearing, and it seemed to make no sense - I'd summon one of the mailboxes from the list, move the mouse over, and *poof*, it would just disappear (but still be listed as open). Very bizarre. I couldn't get the windows to "stick", even after closing and reopening Eudora multiple times. Then I got myself into a very bad situation by telling it to float the mailbox window - and watching that one disappear.

This looked very bad. Then I moved the mouse to another app, and noticed that Windows thought that the center button was down. No amount of toggling fixed that - Windows was just in a bad state. Finally, I rebooted to clear that, and all was well. Really, really weird.

In comments to this post, Chui Tey said something I thought ought to be pushed up:

However, the phenomenal growth of aggregators again underscore the utility of simple, widely accepted formats in delivering tangible benefit to end-users. The world of WS-* in contrast spends 80% of their time making allowance for the 20% of users.

About the size of it...

I just love this quote from an SD Times story on Java platform evolution - Graham Hamilton writes:

We are resisting the temptation to make Java language changes in Mustang. We recognize that James Gosling achieved a genuine masterpiece with the simplicity of the Java language design, and we want to be very cautious in evolving the core language.

But at the same time, languages do need to evolve, and we are exploring a few key changes for the Dolphin release. We’re interested in introducing direct support for XML into the Java language. Many Java developers work with XML, and we’re interested in finding ways of smoothing that integration.

Simplicity? Of Java? Who are these people kidding? Compared to C++, perhaps. I'd guess that the author of this piece has never seen (possibly never even heard of) Smalltalk, Lisp, or Python. Simplicity my foot :)

And "direct support for XML" in the language? You have to love that - after telling us how "simple" Java is, he implicitly admits that it's lacking in power, since they have to extend the language in order to make XML easy to deal with. Hasn't been a problem in Smalltalk - I've had few XML issues in BottomFeeder or Silt. Then again, I wasn't dealing with a language designed by the sort of people who think "final" classes are a good idea, either.

I'm not sure whether this part should make me feel better, or worse:

The Java language is only one of many languages used with the Java platform. As part of Dolphin, we are planning to add a new Java Virtual Machine instruction, which is targeted at so-called “dynamic languages,” such as Groovy or Python. These languages need relatively elaborate mechanisms for executing method calls, and it seems that providing direct virtual machine support will both accelerate execution for these languages and provide final indisputable proof that the Java platform and the JVM are targeted at more than just the Java language.

Elaborate mechanisms? You mean things that VM teams have had in Smalltalk and Lisp VM's for - I don't know - about 2 decades now? It's nice to see that Sun is catching up with the latter part of the 20th century.

SDTimes has a story on IBM's shift of VAST over to Instantiations:

Instantiations characterized the move as a partnership, but an IBM spokesman who asked not to be named said it did not meet IBM’s legal definitions of a partnership and that the agreement was a way to end the life of the product while still providing support for its current customers.

IBM retains ownership of the VisualAge intellectual property, according to Instantiations CEO Mike Taylor. “We’re not buying anything,” he said. “We’re licensing the technology and the right to redistribute it from them. IBM retains intellectual property rights.”

According to Taylor, Instantiations’ corporate history goes back to Smalltalk’s creation. “Our roots go back to the very beginning days of Smalltalk, to 1980 when it was part of Xerox PARC,” he said. The company’s co-founders developed the first commercial version of Smalltalk at Tektronix in 1984, he added.

Hmm - interesting how IBM went out of their way to state that this was a way to "end of life" the product - and if you read the story in SD Times, they emphasized the migration aspect with Synchrony (i.e., the "we send an army of consultants in and convert you into a WebSphere shop path") more than the Instantiations path. It's an interesting story, just looking at the emphasis.

Steve Rubel points to an interesting comment from InfoWorld's Matt McAlister:

"The day InfoWorld's top news RSS feed received more requests than our home page, I started thinking a frightening thought - RSS is doing to the Web today what the Web has been doing to print for the last several years. We have disintermediated our Web site by offering our news in an easier to access format...again."

Well, maybe. What it means is that having a dull, brochure-ware site is no longer enough. You have to syndicate content that will give people a reason to visit whatever you aren't syndicating. If there's nothing there worth reading - you just didn't realize that you were failing before.

The Cincom Smalltalk engineering team has hit another delivery date - the summer release is out. This is a maintenance/bug fix release for both ObjectStudio and VisualWorks - as I've stated before, our current release plan calls for a major release once a year in the winter, with a maintenance release in the summer. If you're a customer or academic user - expect a CD shortly. If you use NC, we should have the new bits ready for download shortly.

Read the "what's new" information here.

Update: The NC site has been updated. If you've already registered, just hit the link you were sent in email and grab the latest. If you haven't downloaded before, register here!

Steve Kelly has some news on the DSM front:

Great news for those who would rather see co-operation than fighting over how best to obtain the benefits of model-driven development: Microsoft's Alan Cameron Wills and I will be running a workshop on Agile Development with Domain-Specific Languages on Monday June 20th at XP 2005 in Sheffield, UK. Alan will be familiar to many as co-author (with Desmond D'Souza) of the Catalysis approach to components in UML, and is now working for Microsoft on their DSL tools. To participate in the workshop, send a page or so about your Agile / DSM experiences to Alan by June 5th - more details on the workshop home page.

James Governor notes that MS is moving more quickly on RSS/syndication than Google is - and that they are being more transparent about it as well:

What is more, this post from Richard McManus is spot on in pointing out that MS is arguably ahead of Google when it comes to using RSS. Google's support for RSS is pretty lame so far, and unlike Microsoft, Google doesn't even have an enterprise business to support and defend. Where do i go to search blogs? Not Google but technorati or blogdigger. Google news doesn't support RSS. I can post blogs to RSS through MSN Spaces just as easily as i can use Blogger, and can find neat hacks in that regard.. Noone has built a vendor supported third party aggregator on top of Gmail (as far as i know), while Newsgator remains a favourite of RSS-hungry Outlook users. And you know what--MS employees actually usefully blog, which is more than you can say for our friends at the Googleplex.

Jon Udell has some thoughts up on dynamic languages:

The notion of custom-built "little languages" goes all the way back to Lisp and Smalltalk, as ultimately everything related to dynamic languages does. It's one of those deeply elegant principles that can take decades to unfold.

Interactivity is another. When I met with Jim Hugunin recently, he told me that when he shows IronPython to folks inside Microsoft, they're most impressed by his ability to wield .NET libraries in an exploratory way from the command line. Who would have thought that the read-eval-print loop would seem like breakthrough technology in 2005?

It's kind of amazing to see simple interactivity still blowing people away. I did a webex demo yesterday, where I spun up a workspace inside the BottomFeeder runtime and proceeded to demonstrate the idea of finding related content - by searching the live data and then slamming the results into the application - here's the code snippet:

| relations items feeds   viewer|
relations := Dictionary new.
relations at: 'aggregator' put: #('rssbandit' 'rss bandit' 'bottomfeeder' 'newsgator' 'feeddemon' 'feed demon').

feeds := RSSFeedManager default getAllMyFeeds.
zooms := Core.List new.
feeds do: [:each |
	items := each allItems.
	items do: [:eachItem |
		relations keysAndValuesDo: [:key :values | | matchOrNil toMatch |
			toMatch := eachItem description.
			toMatch notNil
				ifTrue: [matchOrNil := values detect: [:eachValue | ('*', eachValue, '*') match: toMatch] ifNone: [nil].
				matchOrNil notNil ifTrue: [| zoom |
									zoom := RSSZoomItem new.
									zoom feed: each.
									zoom item: eachItem.
									zooms add: zoom]]]]].

viewer := RSSFeedViewer getSingleInstance.
viewer  zoomFeedItemsView.
viewer  addColumn.
viewer lastStateChange: #goToAllNew.
viewer feedItemsList list: zooms

What that does is define a set of search terms, and then grab all the results together - nothing earth shaking. The part that surprised the demo audience was that I was doing it in a runtime - i.e., a delivered executable. Which leads me to this from Jon:

I do think that professional tools can help dynamic languages consolidate the ground they've been gaining. And to that end, conventional capabilities -- such as suport for testing, debugging, and version control -- will be prerequisites. But dynamic languages really are different in important ways, and their tools should be as well. It's time for some fresh thinking on this topic, and for some new approaches.

Well, the tools aren't only different - they are more flexible. Part of that is the simple fact that in Smalltalk, the environment is built in itself. That gives you a level of flexibility that Iron Python - which has to live within the closed strictures of the CLR - will never achieve. In Smalltalk, I can modify pretty much anything - sure, there's a wall at the VM level, but it's a fairly small one - all the interesting stuff is happening in the image.

This kind of silly putty flexibility gives you a lot of potential - have a look at this screen shot:

That's a browser, running inside the BottomFeeder runtime. I don't normally have source access there, so I'd see decompiled code if I selected a method. However, I could copy the sources for the component in question over and end up with a normal browsing environment if I wanted to.

This is what really separates the mainstream Java/C# stuff from a language like Smalltalk - and it also separates Smaltalk from things like Iron Python, which hae been pounded to fit into the CLR. There's very little separation between runtime and development time, and the developer can shrink that gap down to nothing if he so chooses. In anything JVM or CLR based, you'll always have that separation - always.

Here's an entire family that's been issued tinfoil hats :)

I've updated the server and the posting tool to support hot swapping of CSS files. It's not a complex thing, and I probably should have done this earlier. In any case, there are only two styles available at the moment, but others should be easy enough to gin up if you know CSS. Submissions welcome :)

Dave Winer is still stumbling in the dark, wondering where he might find a clue - this is in reference to some paranoia he's posted about Google along with his obsession with AutoLink:

They can just cut off our air supply. How could they do that. How how how? Well, remember we decided it was okay for them to modify our content way back in 2005. They didn't even have to ask our permission. Not only that, if we said no, they could ignore us. They're just giving the users what they want, and we believed them.

Gee, I sure hope he doesn't use any client side tools that modify content - like, say, ad blockers. Heaven forbid he were to muck with the revenue model of the site's authors, who depend on that advertising revenue. I sure hope he doesn't ever say anything nice about GreaseMonkey either. I certainly hope that he never used a pen or a marker on a textbook, and then passed it on to another student - by gosh the content was marked up in ways that the original author didn't intend! The horror!

I've got an update on Troy's surgery:

Troy’s surgery has completed, and it went very well. They were able to do the surgery lapropscopically, so Troy’s recovery should be shorter and less painful than otherwise feared.

He is still coming out of the anesthesia, and is on a ventilator as a precaution until he is fully awake. But the Dr. said that all went well.

Go Troy!

Early Registration for StS 2005 is ending:

All, Early registration ends May 26 for Smalltalk Solutions 2005. After May 26, the price of registration goes up by $100 USD. Hurry and sign up today before the early registration deadline.

Sign up now to save money!

James Governor thinks the "unthinkable" about Google:

See that mountain in the background that looks so small? Well if the economy suffers a really bad shock, hedge fund collapse, $70 a barrel, or whatever it is, then suddenly the mountain comes to the fore, and the perspective shifts. Could another firm buy Google outright? Lets say the bottom falls out of the advertising market, taking Goog revenues down to $3bn a year (currently doing 1.25 a quarter). Combine that with a major fall out on share price- say take P/E to a more realistic, yet still outrageous, 15. Who could afford $45bn?

Why not buy all the cool technology and brand and fire those expensive PHDs with their 20% own time who don't seem to be generating any cash yet? Why am i thinking about Google in this way? Partly because i have a post on the way about IBM's decision to give Google enterprise props. I am not a financial analyst, as you can doubtless tell from the proceeding commentary. But i just wanted to check myself and think whether Google is buyable/rideable.

Hmm - Which companies have piles of cash available that might be burning holes in their pockets under those circumstances?

James Governor makes some good points about where to go for information on RSS/Syndication/Blogging (etc):

I am skeptical News Corp is doing the right thing in bringing in McKinsey for its attempt to get it. How many McKinsey bloggers can you think of? Better to beg Udell to come along. And far more economical, i should think. You just know McKinsey now has a huge team of new MBAs (trawling the web to "get" RSS and all, probably with sex on the brain). They are being paid by Rupert to find knowledge that others have. Aggregating sure, but very top down and priced accordingly. So why not "view source" instead? Oh yeah- because McKinsey, like a premium beer, is reassuringly expensive. So Where are the McKinsey bloggers?

Pointing to Jon Udell is spot on - he not only follows this area, he lives in this area. Governor's points about equating expense with quality are spot on as well. News Corp is paying filet mignon prices, but is probably getting ground chuck level information...

Funny thing about that last post - immediately after I made it, I ran across this story on Sun's acquisition of Tarantella:

Tarantella Inc. is losing money and customers, but its thin-client technology found a potential savior last week in Sun Microsystems Inc., which is acquiring the Santa Cruz, Calif.-based company for $25 million in cash.

Kind of makes my point, I think. Sun is blinded by their vision of thin network appliances, but the market simply isn't that interested. Full bore PC's are too cheap, and they can function as either network clients or standalone systems.

Frank Hayes of ComputerWorld is debunking Nicholas Carr's latest attempt to define the end of corporate IT. In the process, he makes an interesting side point:

Users are the ones who experiment and create business innovation. So the most important place to put computing, and control of that computing, is in users' hands. Everything else -- networks, data, back-end applications -- is there to support those users. They do corporate computing. We in IT just help.

And if we replace their flexible, too-cheap-to-meter computing with thin clients and a fixed-cost, fixed-services utility, as Carr recommends? IT gains manageability, centralization and higher utilization. Business users lose the ability to innovate

This might be why Sun has had such trouble with the whole network computer thing. CPUs, disk space, and memory have been getting cheaper, so it makes less and less sense to centralize. The only cross current is the difficulty inherent in backing up so much data.

Sean Malloy passes on some news about Dolphin's plans for the product:

Andy Bower mentioned a while back, that Object Arts were toying with the idea of releasing a free version of Dolphin at the same time they released the full version of 6.0.

While they still haven't decided just what version of Dolphin they will be releasing into the wild, but they have now definitely decided that a version of Dolphin Smalltalk, will be available for free under the name of Dolphin Community Edition.

This is pretty exciting news for a couple of reasons. Firstly, Dolphin Smalltalk is the best pure Windows Smalltalk. If you are interested in developing applications on Windows, then it is without a doubt the version to use. Secondly, with a free version of Dolphin, it sets the entry level to a point where people who may not have checked it out before will definitely look into it now.

This is not official news yet. I knew about it for a while, Andy linked to the new Object Arts site in the beta test newsgroup, and it was mentioned on the new site. I asked if I could write something about it, and he didn't say no. So while I don't think that they are going to change their mind about a free version now, they may change their mind about what exactly a free version contains.

Interesting news, and it should help spread the use of Smalltalk.

Tim Bray asks some incisive questions about the terms for Google's AdSense for RSS Feeds:

What’s a Feed User? For a real dose of Fear, Uncertainty, and Doubt, check out the “Terms and Conditions” language. Uncharacteristically for Google, it’s a bad piece of legal drafting; you have to agree that “You will ensure that each feed user complies with, and each feed user’s display of AFF Ads is in compliance with, all of the terms and conditions of the amended Agreement applicable to the Site in the same manner that such terms and conditions apply to You and the Site.”

Huh? What’s a “feed user”? I’m pretty sure it doesn’t mean a person-or-program-reading-the-feed, but I’m also pretty sure that I don’t understand what it is they mean. And until I do, I’m damn well not going to sign up for “You agree you will be responsible and liable for any and all use of the AFF Ads by any feed user and will indemnify Google for any lawsuit or proceeding (a) relating to or arising from any feed user’s use of AFF; (b) relating to or arising from Your failure to ensure any feed user’s compliance with the terms of the amended Agreement; and/or (c) brought by a feed user against Google that arises from or is related to Your provision of AFF Ads to such feed user.”

This has the look of another license released by the lawyers without any "real people" having been consulted.

Let's wish Troy luck and a speedy recovery from his surgery. A friend of ours locally got the same surgery a few months ago, and he looks great - hopefully, things will go as smoothly for Troy.

Darren Oakey, commenting on David's post, has not only drunk the koolaid, he's swimming in it. To wit:

Static typing makes life easier because it SIMPLIFIES the decisions you have to make. Then it makes life orders of magnitude easier because it allows the IDE to provide useful and high quality information about what you want to do - there without changing context, without looking a different IDE. Final classes, checked exceptions make life EASIER because you don't make mistakes. The protect you from both misuse, and the far more common programmers affliction of stupidity - and that protection makes programming EASIER because your program works. First time and every time.

Yep, "final" makes things easier, IF the original designers thought of every possible contingency that I as a developer might ever run across, anytime, anywhere. Static typing makes it so easy that developer never, ever have to use casts. Not once - it simply never happens.

The last part of his comment is the telling part - he wants these things in order to protect the developer from his own stupidity. Just remember that the next time you use one of the mainstream languages - many of the "features" were designed with the idea in mind that you, the developer, are a moron.

As to his last comment, that these things ensure that your program works right the first time and every time? Pardon me while I go have a giggle fit...

Register for Smalltalk Solutions and find out how the chip fab industry works:

Commercial Machine Control using Smalltalk: An experience report from the Semiconductor Industry. experience report Raabe, Thor: Unaxis Wafer Processing Monday 9:15 am to 10 am Abstract: Unaxis has been shipping production equipment for Semiconductor Manufacturing for almost a decade with control systems based on ControlWORKS, a commercial framework written in VisualWorks Smalltalk. This presentation will describe the equipment, its capabilities, the use of Smalltalk and the ControlWORKS framework, and some experiences gathered while developing and supporting these products.

Bio: Thor Raabe is a Lead Software Engineer at Unaxis Wafer Processing where he has been developing semiconductor manufacturing equipment automation software using Smalltalk for 9 years. He has a full range of expertise ranging from troubleshooting hardware and driver code to system architecture and framework development.

See you in Orlando!

Slashdot reports that the MPAA thinks that digital piracy is costing Lucas money:

MPAA President and CEO Dan Glickman: 'There is no better example of how theft dims the magic of the movies for everyone than this report today regarding BitTorrent providing users with illegal copies of Revenge of the Sith. The unfortunate fact is this type of theft happens on a regular basis on peer to peer networks all over the world.'"

Earth to Dan Glickman: Sith has broken the single day record for movie revenues - so far it's raked in $50M. The entire set of films to date has brought in $5.7B in revenue.

So if this is the "best example" of how piracy costs film makers money, I think we can stop worrying. Is Glickman really this stupid? Does he have any idea how much of a moron this makes him appear to be?

David Weinberger lets the Boston Globe (and by extension, the entire MSM) have it with both barrels:

Note to Globe: You, Huffington, Walter Cronkite, the NY Times and the Mayor of Reading are all welcome in our blogosphere. But your concern that your high-toned bigness might just drown out our wee voices is misplaced. The blogosphere isn't a town the professionals can buy up; it's an infinite landscape that will have towns of every sort. We little, irresponsible bloggers are going to continue to find one another and delight in one another. And now and then we're also going to drop in on the upscale respectable towns — well, not the gated ones, of course — and, yes, sometimes we'll be carrying cans of spray paint. But we damn well will not be daunted.

Unlike radio and TV, the net isn't a place that can be bought out by a few gatekeepers - the barrier to publishing has been reduced to a simple desire to be heard. As I said in my last post, the only value-add that the MSM is now capable of adding is in actual, honest to goodness reporting. The kind that costs time and money.

Rogers Cadenhead hits the bullseye on the problems faced by the old line media as they try to adapt to the changed landscape. He first notes that Paul Ford tries to analogize from the cable tv experience - to wit, HBO managed to add value, why can't the Times? To which Rogers points out:

Interesting comparison, but I think it's far easier for HBO to beat network TV than for the Times to beat a horde of free online papers and bloggers.

That's a very good point. Consider cable TV, like HBO - they have more leeway than the free competition, as an artifact of FCC regulation. They can air content that simply can't make its way to the networks. You may agree or disagree with those restrictions, but nevermind - they exist, and they allow HBO (et. al.) to shift the playing field.

Now consider the plight of the Times. Straight news? There's tons of free content out there, from Google News and tons of other sources. Opinion, from their op-ed folks? I can find scads of free opinion pieces across the blogosphere, covering a much more intellectually diverse range than the Times op-ed page allows. What does that leave? It leaves real reporting. It leaves investigative pieces that bloggers don't have the resources to engage in, and that cable news nets won't air, for the most part.

Can the Times add enough of that to justify their subscription costs? I don't know, but the legacy of the Jayson Blair incident certainly doesn't help their cause. It's going to be a tough row to hoe, that's for sure.

I'm slightly more than halfway through "The Birth of the Modern", by Paul Johnson. It's a social/cultural/scientific history of the period between 1815-1830, when Johnson posits that the modern West was born. It's a fascinating book - kind of like the old PBS series "Connections" in the way it bounces from topic to topic, but in a more restricted time interval. Here's where the title of this post comes from - in a section on the rise of the Scientific method, Johnson writes:

In the years after Waterloo, scientific invention was of passionate interest to a rapidly expanding British and international public. That was the most important new factor. But it was still possible for a moderately well educated man or even a woman - a manual on chemistry was specifically written to appeal to ladies - to grasp the latest scientific developments. Indeed, an empiric engineer like Stephenson, who had no schooling, worked at the frontiers of technology alongside scientists like Davy. Physics and chemistry, science and engineering, literature and philosophy, art and industrial design, theory and practice - all constituted a continuum of knowledge and skill, within which men roamed freely. The notion of separate, compartmentalized "disciplines", later imposed by universities, did not yet exist.

Before someone gets their shorts in a bunch over the way he wrote about women there, recall that this is the early 19th century being discussed, not the modern era. The fascinating bit to me is that anyone with enough sense and curiousity could work at the bleeding edge back then - if modern knowledge is a staircase, people were still standing at the lower landing, peering up. It's certainly not like that now.

That's why the 19th century has such a romantic allure to it - it was, by and large, a time of piece in the West. Certainly there were large disruptions - the Civil War in the US, the revolutions (mostly abortive) of 1848 in Europe, the Franco-Prussian war. None of them swept the world though - there was nothing as huge as the wars or Napolean, or as large as WWI became. That meant that people were free to spend time, energy, and money on scientific and cultural pursuits. That's what happened.

There's a notion that science is advancing faster than ever before, but it's really not the case - the technologies introduced through the course of the 19th century and into the early 20th were far, far more disruptive. Consider communications technology, for instance. In reading about the negotiations at the Congress of Vienna, it was clear that the world hadn't changed yet - the ministers sent were mostly on their own, as dispatches from Castlereigh (of the UK) back to London took weeks going back and forth. Skip a few years forward and you had the telegraph - which brought instantaneous communication to the world. In 1815, it took months for the news of the volcanic eruption (which caused "the year without a summer") to reach Europe and the US. By 1883, when Krakatoa went up, it took a day or so. It's faster now, but not dramatically so.

Look at transport - with bad roads and horse drawn conveyances, it took days to travel from New York to Albany. With the introduction of the railroad, that dropped to hours. Sure, planes sped that up even more, and removed the need for the high cost railbed. The disruptive change in circumstances came with the rails though. Word of a news event could travel instantly (telegraph), and people could journey to and fro in hours, or days - instead of weeks and months. The coming of the steamship in the latter half of the century collapsed sea travel in the same way.

Now consider the changes we've faced since the 1960s, for instance. Computers and the internet? Well, computers existed back in the 60's, and we (at least in the West) were already attuned to the idea of changing technology. What we got was better and faster, not completely new. The PC and the laptop were incremental improvements, not wholly disruptive events. In the grand scheme of things, the telegraph was a bigger shock to the system than the internet was.

In the whole hullabaloo about Ballmer's alleged RSS comments (the original poster makes it clear that what he posted is not a verbatim transcript), I missed the most relevant tidbit - one of my readers pointed this out in email:

A. We believe RSS is important and will be around for a while but it is not going to change the world. It is a little too simple, that is also the reason everyone’s using it. We are working on more existing powerful stuff, around XML/web services [sic] that will address many issues beyond RSS. RSS will be around, but whatever we are working next will be cooler and more prevalent.

Emphasis added by me, btw. It's a little too simple? Right there, I think this speaks volumes about the corporate culture at Microsoft (not that it's absent elsewhere) - simplicity is bad - complexity means we accomplished something. Complexity means it's powerful.

In a nutshell, that's what's wrong with the IT sector as a whole.

Well, it seems to be time for my weekly log report - here are the results for the last week:

Platform	BottomFeeder Downloads
Windows	775
Mac 8/9	480
HPUX	349
Sources	335
Linux x86	318
Mac X	241
Update	133
CE ARM	122
Windows98/ME	62
Linux Sparc	21
Solaris	20
AIX	14
Linux PPC	9
SGI	3
ADUX	1
Source Script	1

That's a total of 2884 over the last 7 days, for an average of about 412 a day - not bad, seems to be holding steady. I still don't get those HPUX numbers :) Let's have a look at the feed access:

Tool	Percentage of Accesses
Mozilla	20.5%
BottomFeeder	20.5%
Other	21%
Net News Wire	14.9%
NewsGator	4.4%
SharpReader	4%
BlogLines	3.9%
Internet Explorer	2.8%
Planet Smalltalk	2%
Liferea	1.8%
Feed Demon	1.6%
RSS Bandit	1.1%
JetBrains	1.1%
Feed Reader	1%
Shrook	1%
Java	1%

Those numbers don't look too different from the last log check. Finally, a look at the straight browser accesses to the html pages:

Tool	Percentage of Accesses
Mozilla	42.2%
Internet Explorer	31.8%
Other	15.0%
Java	3.5%
BottomFeeder	2.8%
Net News Wire	2.7%
BlogLines	1%
Opera	1%

Looks about the same as last week, although the aggregator accesses are a bit higher. Then again, the tested interval is shorter.

I was going to see The new Star Wars flick this afternoon - but the rain/wind storm knocked out power in one place in Columbia - the shopping mall where the theater we went to is! So instead, we had lunch (thus the title :) )

One of the cool things about BottomFeeder is that it's wide open to ad-hoc extensibility. I made a post about this 10 days ago - finding related content. This morning I thought I'd do a short screencast on the topic. The quality of this one might not be as good as the last few, but it's way, way smaller - the wmv file comes in just under 2 MB for 3 1/2 minutes. I'll muck around with the audio and video settings and see what I can do with the output in the future, but it here it is.

Here's an interesting tidbit from an impromptu interview with Steve Ballmer of Microsoft:

Q1. How important is RSS? A fad, important, huge or will replace the web/html dominance of the internet?

A. We believe RSS is important and will be around for a while but it is not going to change the world. It is a little too simple, that is also the reason everyone’s using it. We are working on more existing powerful stuff, around XML/web services [sic] that will address many issues beyond RSS. RSS will be around, but whatever we are working next will be cooler and more prevelant.

This touched off a mini-flurry of posting - Dave Winer assumes that it means MS is going to try to bury RSS. Scoble responded with a "look at all the cool RSS stuff we're doing!" post. Steve Rubel noticed the whole thing as well.

Here's the thing - all these guys assume way, way too much. You want to consider what kind of answer you would get had Ballmer been asked "Compare and contrast RSS and Web Services?" He's not a technical guy. He's a manager who was technical a long, long time ago, and I'd bet good money that his grasp of RSS and WS* stuff is limited at best. All this really demonstrates is that Ballmer should defer most of these sorts of questions down to knowedgeable staff instead of trying to answer them.

I've always wanted a Smalltalk balloon, and now I've got one!

Thanks to Jason Jones at STIC!

Ted Leung has a good observation about the Star Wars flick (which I haven't seen yet):

As for the movie itself, it's hard to say. Both the beginning and end were predetermined, so there was a limit to what could be done. Certainly there was plenty of action, and most of that was pretty good. The dialog was pretty bad, as expected. It certainly wasn't as bad as Episodes I or II, but I'm having trouble deciding whether it was good enough to edge out Return of the Jedi. Regardless, after 28 years, Star Wars is done.

The lack of any real mystery in where the film will end up certainly does limit the attraction somewhat. It comes with the territory, having been set in the past of the "first" three films. To some extent, that was the problem "Enterprise" had as well - you knew that it had to end with the founding of the Federation, just like you know that "Sith" has to end with the rise of Vader and the Empire.

Having said all that, I'm sure I'll go anyway.

You can register for the Smalltalk Solutions tutorials now:

Tutorials are now available for signup at:

http://www.smalltalksolutions.com/registration2005.htm

Each tutorial is $100 USD

Tutorials available are

• A Seaside tutorial by David Shaffer
• Web Services by Allen Davis
• Domain-Driven Design by Eric Evans and Ralph Johnson
• Smalltalk Garbage Collectors by John McIntosh

Early registration for the conference ends May 26. After that, conference registration increases by $100 USD. Sign up today at http://www.smalltalksolutions.com/registration2005.htm

See you in Orlando!

I'm not sure that this is what the Discovery Science channel was going for - they were doing a segment on the dangers of driving while tired, and this is what popped up on screen - if it's not clear, that smoky area is a nasty car crash. This happens as the show - "Understanding" (an episode on sleep) was talking about how many people die in car crashes due to falling asleep at the wheel. Then we get this:

If you can't read the text in orange, it says: "They've gone places we've only dreamt of... capturing worlds beyond our imagination". Whoo boy :/

The process we used to get this image to my blog was a story in itself. First, my wife captured a few minutes of the show in our ReplayTV. Then we used DVArchive version 2.1 to download the show to her PC. Then she played the show on the PC until the relevant segment (above), and screen captured it. From there it went to my PC, where I cropped the rest of the screen, resized the image, and sharpened the text rendering a bit. Whew!

As per usual, if it's not Dave's idea, it's no good. Oddly enough, people differ in their viewing preferences - esle we wouldn't need more than one TV channel...

Nu Cardboard notes that even with IGB of RAM in their PC's, the Java apps they run eat the machines alive. The Eclipse IDE alone chews 300MB!

Now, contrast that with the Smalltalk system I use:

• A full development image running the blog server, the survey application, and a few other things: 142 MB of RAM. That's a full bore development image, btw - I have full access to all development artifacts there, on the remote server in Cincinnati
• The test server, which I run on a PII 400 - 63 MB. Go ahead, try running Eclipse on a PII 400.
• My BottomFeeder runtime application, with most of the development tools retained (I can open workspaces and write Smalltalk against the running application!): 110 MB

I think those guys made a sub-optimal choice in terms of development/deployment tools. The development image I use to work on BottomFeeder only chews 56 MB, and that's with all the code for it loaded into memory. The bottom line is, those of you who still think Smalltalk is "bloated" - you need to look again. There's bloat alright, but it's not over here...

Steve Rubel reports that BlogLines has big plans:

The CEO of Bloglines (now a division of AskJeeves) says that his company will release a blog search engine this summer which will surpass the likes of Technorati, Feedster, and PubSub. "The challenge," he says, "is to create world-class blog search, which we don't think exists now."

AskJeeves knew that direct competition with Google wasn't working out, and figured that a sidestep into a related field would be a better tack. Now let's see if they can be faster than PubSub, arguably the fastest search/match system for blogs out there right now.

I really, really hope they don't let Joel anywhere near development tools anymore. He made asinine comments about exceptions awhile back; now he's decided to remove all doubts about his lack of comprehension:

Now, when I’m writing a dinky script to gather up a bunch of data and print it once a day, heck yeah, exceptions are great. I like nothing more than to ignore all possible wrong things that can happen and just wrap up the whole damn program in a big ol’ try/catch that emails me if anything ever goes wrong. Exceptions are fine for quick-and-dirty code, for scripts, and for code that is neither mission critical nor life-sustaining. But if you’re writing an operating system, or a nuclear power plant, or the software to control a high speed circular saw used in open heart surgery, exceptions are extremely dangerous.

I know people will assume that I’m a lame programmer for failing to understand exceptions properly and failing to understand all the ways they can improve my life if only I was willing to let exceptions into my heart, but, too bad. The way to write really reliable code is to try to use simple tools that take into account typical human frailty, not complex tools with hidden side effects and leaky abstractions that assume an infallible programmer.

That's right Joel - we should always return error codes instead, passing them up through 10, 15, 20 (insert your own value here) calls until we get back to the point in the code where the problem can be handled. All because Joel is deluded, and thinks that an exception handler is the same as a goto.

In his preferred world, when a problem crops up deep in the bowels of the app, how does he propose to deal with it? Damned if I know; I'm not sure he knows either. I'd go through all this again, but I already did - read this, and this.

I'll make one other point - in Smalltalk, you can resume execution from the point where things blew up - because the stack hasn't been tossed by the time you get to the exception. Maybe what Joel needs to do is learn a programming language that doesn't suck.

Please, take his code editor away. Quickly.

Blaine Buxton explains why resumable exceptions are a good thing:

Time for another "this is why Smalltalk is cool" post, but this one also holds true for Ruby And Lisp as well. So, it's a "why Smalltalk, Ruby, and Lisp kicks mucho booty" so to speak. OK, enough of the back patting and let's get down to business. Today's topic is resumable exceptions. It has a nice geeky ring to it doesn't it? The first thing you might ask yourself is, "Why in the world would I want to resume an exception? It's an exception! Dead programs tell no tales!" True, true. Normally, you want an exception to send your program down in flames because you had a mechanical glitch that you didn't expect. Better stop everything before the propeller goes slashing through your data unkindly! But, what if we had exceptions that were good that could notify us of potential bad things or even enumerate potential bad things?

I use resumable exceptions quite a bit in BottomFeeder - they allowed me to create a customized RSS/Atom handler that could deal with many of the trivial issues in feeds (like bad characters) without having to create more own "tag soup" parser. The lack of them in the maintream languages (Java, C++, C#) explains why every time I bring this up in a forum with people involved in the syndication space, they assume that I had to roll my own regex based tag soup parser. But hey - all those extra libraries must be making them more productive... somehow.

The next time someone in management asks "why do you spend time monitoring the blogosphere?", point them to this post by Steve Rubel. The next thing you'll want to figure out is "what would our response be in a similar situation?"

Is Gosling really this stupid?

The "clear need" that Magnusson cites is anything but clear to Gosling, who says Sun has received negative response from the enterprise development community regarding the idea of open-source Java. "We've got several thousand man-years of engineering in [Java], and we hear very strongly that if this thing turned into an open source project—where just any old person could check in stuff—they'd all freak. They'd all go screaming into the hills."

I don't know James - has it been a problem for NetBeans? Is the Apache project in chaos? Is Eclipse? If you don't want to open source Java, that's fine, and believe me, I'd understand. What I don't get is why you have to make crap up instead of just saying no.

Schilling is still recovering from that hack job he agreed to during last year's ALCS - and while it got the Sox the AL championship and the Series Rings, I think it may have cost Schilling the balance of his career. I seem to recall mentioning that...

Rob Fahrni asks:

The question is, how does an old dog such as myself get immersed in the gospel of yet another language? You can say all you want about Microsoft but you have to admit their developer tools are solid if nothing else. I know IronPython exists for the .Net framework but what about Ruby and Smalltalk implementations? I'm spoiled to Visual Studio.Net, it's so nicely integrated, and just works. I've really come to appreciate VS.Net now that I'm spending my days on a Linux box, you have no idea how pathetic the tools are on Linux.

Well, here's what I used to do, fwiw: I'd take a problem I had solved in my first programming language (Basic) and write the same application in the new language. That way, I wasn't trying to understand the domain problem, I was just learning the new language. It wasn't a hard problem it was a manual cryptogram solver (for the puzzles that still appear in some newspapers). I wrote that in Basic, in UCSD Pascal, in a proprietary language at the DoD, in C, and finally in Smalltalk. I stayed in Smalltalk after I finished the problem in less time than it had taken me to go over syntax in the other languages I had learned :)

As to the question about .NET integration - there are no shipping Smalltalks on that platform - a large part of the problem is that the CLR just isn't ready for a language like Smalltalk (at least not yet). As to tools sucking on Linux - that's not true if you use something like VisualWorks - which is binary portable across every platform we support :) I do my BottomFeeder work mostly on Windows, but I do all the blog server development on Linux - and on an old PII 400! Try running any of the supposedly "modern" development systems on that :)

Lenny Hoffman, Objectivity engineer. Current query support added in 1992, and hasn't evolved a lot. Not initially a problem - most people were using the product as a persistent store. Customers who need query support either built their own or ran back to an RDBMS. The issue - sheer inertia held this areas back.

The big problem is that data size is growing. There's a growing demand from their customers for out of the box query support - they want the scalability that they've come to enjoy with the OODB, but with the kind of query support they know is available with an RDBMS.

• Higher Fidelity - relationship properties as query values, Set qualifiers, path based queries
• More customizable - Application defined calculated values, Application defined indices
• Better performance - primarily optimization
• Open - publish a public OQL with an independent predicate equation tree. Add a listener interface for monitoring, logging, and tuning

[ed] - interesting bit about OQL - will that be truly open, such that other ODB vendors could adopt it? Or is open in this sense just about having a defined API?

The Path query support is somewhat inspired by XPath, but without the reliance on an XML structure. This looks a lot like what you would do with a Smalltalk collection and one of #select:, #collect:, #reject:, or #detect:.

Calculated Values - this is what you get with Gemstone by having the code execute in the database instead of in the client, or in an RDBMS via stored procedures. With Indexing, the big changes are extensibility and the addition of a public API.

In summary - this is an enhancement to the existing query engine by opening it up and making it easier to access and extend. Another note here - this is not part of the current release. In answer toa question, it's not clear which release it will be part of.

Today's keynote is being given by an FBI CTO who spent most of his career at NSA - interesting cross over. He made an interesting observation about how line workers view updates from IT - they fear that new technology coming from IT will be less capable than what they have now. That's hardly unique to FBI - and it's part of a large scale issue in IT shops: there's a lot more slavish following of analysts (especially the useless ones) and fads than there is examination of actual needs.

One item I noticed he hit on as being important to him - SOA. The thing about SOA that makes me skeptical is that it means all things to all people. It's like OO 15 years ago, or the visions being pushed by the OMG a decade ago. Universal answers never are - but C level IT folks always seem to be willing to drink the koolaid.

Oof. In a discussion of network security we just got treated to "buzzword bingo", where the buzzwords are all security agency specific. There's a real issue afoot there though - in any secure environment, how do you make data available to people in an appropriate manner? For instance - some data might need to be shielded for reasons having to do with trial rules. Others might have to be shielded based on which foreign governments are or are not allowed to have access. That all sounds like "why do I care?" for business folks but - with the advent of SarbOx rules, it actually has deeper meaning than a lot of us might like.

Here's another issue that will resonate with business people - the simple volume of structured (and unstructured) data washing though systems. Consider a product marketer or product manager - how the heck do you figure out what the competition is up to? For that matter, how do you figure out who the competition even is? Unlike the FBI/NSA problems, it's nearly all open data - but the fact that it can be found doesn't mean that it will be found. And here's where the FBI guy relates it to his problems:

When the military captured Hussein, they used social (or network) analysis to do so. The tools were pretty simple - large sheets of paper and markers. Now consider a hostage taking (something the FBI deals with). In such cases, they don't have a long time to find the person - it's literally life and death, and time matters a lot. Being able to drive that kind of analysis quickly using decent tools would help them a lot. To relate this back to what seems to be a trivial issue (at least when compared to kidnapping), consider a product team trying to determine which of a set of desirable features to implement for the next release of a product. There may be (say) 10 possible features of interest, but - given the size and capability of the team, combined with the desired delivery schedule, only (say) 3 can be delivered.

How do you make those calls? Market research to determine sales impact? Interviewing existing customers in order to extrapolate general market demand? What about extending the delivery time out so that more could be done? What about hiring more staff so that we could accomplish more? These are all calls that are made with insufficient data, and they need to be made decisively. In my work, I make those sorts of calls all the time, and I'm never entirely certain that I've made the right ones.

Interesting tidbit just came out at the tail end of the talk I just made it to (blasted beltway traffic!) - Objectivity now has a Python binding. That's pretty cool - they also support Smalltalk, of course. This means that you can be productive with Objectivity, instead of falling down into the endless complexity provided by C++ and Java. The presenter pointed out how much more difficult it is to do prototyping against Objectivity with C++ or Java. They plan to release official support for Python later this year - the binding itself has been around since 2000.

Dare makes some very good points about the problem of synchronization - he outlined the problems with the current BlogLines API earlier:

most popular desktop aggregators on the Windows platform will now have a richer synchronization story with its most notable competitor. It also puts pressure on other desktop aggregators to figure out a strategy for their own synchronization stories. For example, I had planned to add support for synchronizing with both services in the Nightcrawler release of RSS Bandit but now it is unclear whether there is any incentive for Newsgator to provide an API for other RSS readers. Combining this with the fact that the Bloglines API isn't terribly rich means I'm between a rock and a hard place when it comes to providing synchronization with a web-based aggregator in the next version of RSS Bandit.

Ditto for BottomFeeder. Right now, BottomFeeder can synchronize itself with another running instance of BottomFeeder over HTTP, or via file import (i.e., you export a synch file from one copy, and then import it into another). Until there's a useful server based API, I don't see anything happening here on my end. I don't really see what incentive the NewsGator guys have for playing nice, either. That's not a shot at them - far from it. I just don't see why they would worry about it...

Steve Rubel points out an evolving phenomenon:

People now put the same degree of trust in me (and other reputable bloggers) that they might normally only reserve for analysts and journalists. There's a big difference between these influencers, however, and bloggers. With press/analysts you have a safety net. If they leak, you have options for recourse. With bloggers you really don't.

Well, I'm not clear on what safetly net exists with analysts either (unless you have a contract with them). Ultimately, you're making a trust call, and the person you confide in is going to either be trusted in the future - or not - depending on their behavior.

The more interesting aspect of this is that you could be "on the record" at any time now - whether you are giving a formal presentation/speech or not. That's going to be a rather large change as realization starts settling in.

I'll be at the Objectivity show most of the day - make sure you get your Coding Contest submissions in before 9 AM!

Darren Hobbs wonders about the new method of class creation in VisualWorks (new since 1999 - which is to say, not that new :) ):

Hmm. VisualWorks has namespaces, which Smalltalk-80 does not, and they seem to have changed the way subclasses are defined. According to my copy of the purple book, a subclass is created by sending a class the message 'subclass:'. It should be possible to implement a version of subclass in the Creature class that adds the extra code.

However, VisualWorks wants me to send 'defineClass:' to a namespace. Problem with that is I want to change the way subclasses of a particular class are defined. If I change the code in the NameSpace class that will affect every class that is ever created thereafter.

How is this different than before? Prior to namespaces, if I modified the class creation method in class Behavior, I ran into the same issue. Nothing but the message recipient has changed, IMHO...