The mod_pubsub folks are putting forth mod_pubsub as a solution for RSS/Atom item distribution and PUT/WebDAV caching. Another piece of tehcnology to put on my list to look into.

Todays quickies are about paying for open source / the commoditization of software.

• Rod Waldhoff thinks that The Silent Majority Pays for Open Source.
• Matt Langham compares the commoditization of software with the commoditization of flying.
• The ServerSide is hosting a thread on the who pays for open source topic.

Kevin Altis pointed out some interesting text in the Python 2.3 press release:

"The combination of the open source Unix-based core of Mac OS X running on PowerBook G4 high-performance portables has attracted a large number of developers using open source scripting languages like Python," said Bud Tribble, Apple's vice president of Software Technology. "Python 2.3 provides greatly improved support for existing Mac OS X users, and with the upcoming release of Panther, Apple will provide Python 2.3 developers direct access to APIs for the PDF-based Quartz graphics engine and QuickTime image formats."

The Mac OS buzz just keeps getting louder. Too bad the market share keeps getting smaller.

• Steve Gillmor and Rajesh Jain both agree that RSS means new life for fat clients. Rajesh has picked up on the Chandler as microcontent client meme.
• Via the Floating Atoll (sorry, I couldn't find your name) another good source for an RSS feed: PhD dissertations. I'd expand that to all theses and technical reports (or the equivalent depending on the field). That kind of leads to the next item.
• Tom Coates observes that daypop and blogdex have been overwhelmed by political posts from the U.S. and calls for balkanizing them, so that there is some opportunity for other ideas to be heard and found. There's a role here for topics and categories. People have been talking about topical weblogs but I'm not sure I'm ready to go that far. I am more than happy to have topical aggregators like JavaBlogs or SeaBlogs, etc. where I can go to trawl for new bits.

Mark posted in response to my suggestion to merge Subversion with his client side WebDAV cache. I totally agree that the cache should be a general DAV+DeltaV cache. I just think that some of the Subversion / DeltaV stuff may help in resolving merge conflicts when you try to sync the cache after being disconnected for a while.

I was very happy to read that Dan Sugalski decided to go with continuation passing style (CPS) as the internal representation for the Parrot VM. But things that are internal could be made visible, which would make Parrot a good candidate for hosting Scheme or Stackless Python. Very interesting.

Reading Dan's post rang a bell in my head regarding CPS and Static Single Assignment (SSA) form. When I left graduate school. SSA was becoming the IR of choice for compiler optimizations. Richard Kelsey wrote a paper regarding A Correspondence between Continuation-Passing Style and Static Single Assignment, which showed how to convert programs in CPS form to SSA form and the reverse. There are few CPS programs that cannot be converted to SSA, but they are not generated by the usual Lisp/Functional language to CPS conversion algorithm.

Werner Vogels is out at Microsoft for the Microsoft Faculty Summit. One of the things that he discovered is that X# is based on Meijer and Schulte's paper Unifying Tables, Objects, and Documents.

It turns out that some folks out there have been working on time-warp debuggging for Jave. There's Bil Lewis' GPL'ed Omniscient debugger, as well as Visicomp's " RetroVue, the Total Recall Debugger", which is available for $995.

Someone at Eclipse should start paying attention...

Ben Hyde has been a source of insightful thinking for me. His latest noodler is all about the identity problem.

The posts about text summarization have stirred up some good activity. Kellan Elliott-McCrea has done a nice comparison of the Mac OS X summarizer, libots, and Classifier4J. Among his discoveries: there is no system API to the Mac OS X summarizer. Surely this is a sick joke. One of the supposedly wonderful things about closed source, Objective-C based, heaven on earth, Cocoa nee NeXTSTEP was the ability to provide system wide services in a cool way. Boo.

Eager to get a better comparison, Nick Lothian the author of Classifier4J has put up a web app so that people can test the quality of Classifier4J's summaries.

For a long time I've wanted a text summarizer that I could use as a system service. It looks like there's some healthy incentive for the authors of these three systems to keep improving them. Information distillation, here we come.

Some tools related quickies:

• Today is a Mozilla.org day with both Thunderbird 0.1 and Firebird 0.6.1 being released. I installed Thunderbird and got it working against my existing Mail directories from my Mozilla profile. This was actually somewhat urgent because Mozilla 1.4 has a bad GDI leak on Windows, so after some time, the browser and mail reader go belly up. Plus when you open too many tabs in the browser then you can't read mail because the whole suite is locked up. I spent too much time playing with Thunderbird, so Firebird will have to wait, probably till 0.7, because I don't want to hassle with all the extensions and whatever.
• A Mozilla and Thunderbird related tool is Enigmail which is a nice extension for both Mozilla and Thunderbird that lets you use GNU PG. On Windows you can couple this with the Windows binaries for GNU PG at nullify, and you have an entirely open source OpenPGP mail system. I even brought my PGP 8.0 keyrings and everything over. ASF people take note.
• I like thinking tools like outliners. It seems that FreeMind, a Java mindmapping tool has taken some big steps since I looked at it last, especially since they are trying to compare themselves to MindManager, which I've been thinking about buying.
• BEA has a new extensible compiler framework called Javelin. It looks cool but it also looks like a binary license. Fortunately, I know a very extensible Java compiler framework. It's called Eclipse.
• Which brings us to our last tool quickie of the day, which is also related to BEA, but on a happier note. Since it looks like XMLBeans is going to incubated at the ASF, I'm paying more attention to XML databinding than I have in a while. So the Avaya workshop on XML and Data Binding caught my eye.

Via mamamusings comes the Blog Change Bot. Before I really got into blogging I subscribed to Simon Fell's Blog Toaster, which seems to be down or full.

Mark Nottingham thinks that using WebDAV and a client cache in front of the WebDAV server could solve some of the offline/synchronization problems posed by mobile computers. He notes that the problems that arise in this approach have to do with versioning and items changing underneath. I think that if you combine the client cache with something like Subversion, which is already based on WebDAV and DeltaV, you might have an interesting solution. The problem of resources changing behind your back when you also have changes is what Subversion is all about. So "all we have to do" is build an offline aware client side cache for Subversion...

The ServerSide's interview with Gregor Kiczales had an interesting tidbit tucked into it: a presentation by Ivar Jacobsen that shows how use cases have cross cutting concerns and are thereby amenable to AOP. This is good because the AOP people need to show cases beyond the systems level crosscutting concerns. Jacobsen's presentation is a good step in that direction.

So here I am linking Robert X. Cringely, who has proposed a wild but interesting scheme to change the game in the music business. I'm not entirely sure that it would work, but it's good to see "lateral solutions" for some of the issues facing us today. I wonder if there's an analog of this for software.

Matthew Thomas posts about various weblog tools and their permalinks. He points out all sorts of "cruft" in the various permalinks.

I was trying to get to something like this for my pyblosxom blog, but it didn't work quite right. So you'll see that my permalinks are in YYYY/MM/DD#NN format, where NN is a number identifying a post. But the #NN part is bad, and I'm probably going to change that at some point.

Nick Lothian tracked back with his checkin of the latest version of Classifier4J. This is definitely worth looking into for Java projects. libots is still interesting because it's in C and could be wrapped for Python or Ruby.

Nick also has a cool hover trick for the background of the blog entry that your mouse is over.

Via Luke Hutteman:

Jason Brome has cooked up a cool way to make it easier to subscribe to news feeds. I hope that FeedDemon can make it onto the list soon.

Adam Bosworth left Microsoft, founded Crossgain, and is now Chief Architect at BEA. He's just started up his blog and is already in the thick of it. FWIW, I agree that disconnected or semi-connected applications are going to become more common. If you've read the recent articles by Tim O'Reilly and Ray Ozzie, you won't be surprised at this. I do think that both Adam and Tim Bray are not quite on in their assessment of "rich clients". But I don't think of it in terms of UI complexity / simplicity or the weight of the applications. I think that its interesting for people to mix data that they get via the network with data that they have on their local machine. The problem with the browser isn't necessarily that the UI is not rich, it's that its mostly a world unto itself. What we need next is a way to break down the wall between data that we got from the network and local data. And that's why offline is important.

In any case, I'm glad to see Adam on line, he's a deep thinker. I personally owe him a debt that he probably knows nothing about, since we've never met. Back in 1998 when I was working at IBM, and Adam was at Microsoft, we both attended an IBM internal XML summit being held at the IBM Almaden Research center. I was there because the group that I was in was about to start working on XML. Adam was there to talk about XML. I was shocked as he stood up and proceeded to tell the IBM folks exactly what they needed to do in order to capitalize on XML. He was right, and the technical arguments and vision that he set out in his presentation played a big role in my getting excited about our assignment to work on XML. That led to our work on XML4J, and subsequently Xerces-J, which led to my involvement with the ASF. So thanks Adam!

John Robb linked to DEVONthink which is a free form information manager for MacOS X. It takes a less structured approach than Chandler is trying to take. It looks like you just dump all your information in there and turn it's recognizers loose and it sorts it all out for you.

One thing that I noticed while reading the pages is that Mac OS X has a text summarization service built in. I've been looking for something like that for a long time. I know that the ATG (Advanced Technology Group at Apple) had one of these lying around. This is a great thing to have as a system service. If one of you MacOS X hackers can confirm this, that would be great.

In the process of reminiscing about this, I decided to do some Googling. It turns out that the Open Text Summarization library being used in AbiWord is now up on SourceForge.