Month: November 2002

Kenneth Hunt points out that RAM is falling again.

Ovidiu Predescu writes about an oh-so-cool J2SE -> J2ME shrinker:

Sylvain Wallez’s J-Fit seems to be an amazing tool. It essentially rewrites the code to remove debugging statements, rewrites the code to work with J2ME which lacks important functionality. Using this tool, Sylvain was able to fit Cocoon in a mere 1.1Mb jar file, and have it running on a PowerPC 50MHz with 32Mb of RAM. Apparently the tools is also very small, about 10 classes each with less than 500 lines of code. All the XSP pages are precompiled, so there’s no compilation happening. The result is that requests are served in less than 300ms. This includes time spent by Saxon, the XSLT processor, whose results are not cached at all. With precompiled pipelines, XSLT would no longer be necessary which would speed things up quite a bit.

Hopefully he’ll release the tool as open-source, so we can all make benefit from it.

I’ll be attending at least one day of LinuxWorld this year.  I currently have an expo pass, but might sign up for a single day of sessions.  Is anyone out there reading this going to be attending?  If so, let me know.

LinuxToday reports that Samba 2.2.7, which looks like a security fix.

Various bits of Mono news this evening:

• Sebastien has got DSA and RSA signatures working as well as RSA encryption. We now distribute Chew Keong TAN’s BigInteger classes.
• Brian has contributed a System.Data multiplexor in Mono, it can be found in the Mono.Data assembly. The details of this new technology are here. It works in Mono and the .NET Framework.
• Larry O’Brien has announced the candidate book for Thinking in C#. The book is Mono-friendly.
• Dan Morgan has implemented a DataGrid widget for Gtk#, you can see Windows screenshots for it here and here.
• Slides from the Mono developers for the .NET ONE conference are available now:
• Mono Keynote presentation
• Hosting the Mono Runtime
  The simple embedding of Mono in Perl is available here
• The Mono JIT compiler
• Mono C# Compiler Overview
• A couple of other presentations from Miguel’s trip to Europe are available here in Open Office file format.

And Yes, playing with OpenMosix is a form of structured procrastination.

It looks like Java programs utilizing “Green Threads,” which was the default before JDK1.2/1.3 will migrate to other nodes, while “Native Threads” will not.  I’ll have to see if there is a compiler or java runtime flag that will allow me to force green threads.

How about a Gump (or similar Ant) build over an openmosix cluster?  That’s what I’m thinking of…

Preliminary searching reveals that green threads are no longer available (at all I think) in the 1.3/1.4 JVM:

Version 1.1 is based on green threads and won’t be covered here. Green threads are simulated threads within the VM and were used prior to going to a native OS threading model in 1.2 and beyond. Green threads may have had an advantage on Linux at one point (since you don’t have to spawn a process for each native thread), but VM technology has advanced significantly since version 1.1 and any benefit green threads had in the past is erased by the performance increases over the years.

I noticed that every once in awhile, openmosix would suck all of the bandwidth on my D-Link DES-3205 switch for a split second every once in awhile.  If I end up adding any additional nodes, I’ll probably migrate the openmosix machines to their own DES-3205 so that they don’t cramp my AA:O.

This is probably okay, as I need to bring another DES-3205 online.  By the way, if you’re looking for high-performance managed 10/100 switched networking, adn don’t mind the noise, they can be found on ebay for as little as $25-30 plus shipping.  They actually run on AMD K6-II’s, which I found amusing.  They are upgradable with modules, but I’ve never seen one in the wild.

I upgraded two machines at home (Queue and Borg) to the latest OpenMosix kernel, and managed to get the openmosix userland stuff up and running after I googled my hostname problem.  I ended up doing most of the work while at school today.  I’ll have to run some stress tests at some point and try to find something useful to do with it.  My next cluster-related task is to try to get openmosix running on my Red Hat 8.0 Via Eden board so I can run openmosixview and 3dmosmon running, which would be truly cool.

So far I do know that if I start a bunch of stuff on one machine, the load on the other shoots up, which is probably a good sign 🙂

Today’s project was brought to you by the word college, the letters SSH, and the program Putty.

Clemens Vasters weighs in on the current WSDL is too <…> debate:

Hard or not hard — can we agree on “It’s just not enough” ? 🙂 My main problem with WSDL is that it tries to do 2 things (message contract and transport mapping), while it should do 3 things (message contract, service contract and transport mapping), hnowever at the same time, one thing (WSDL) shouldn’t do all these 3 things altogether but leave them two 3 separate things: A message contract definition language (defines soap:Body content), a service contract definition language (soap:Header) and a “web services binding language” that maps messages combined with services to transports.

I definately would not mind three straightforward description languages (MCDL, SCDL, and WSBL) if they were easy to use (with boilderplates or whizzy programs), solved more problems than WSDL does, and were easy to use.  I think Greg Reinacker’s earlier point that schemas are important is true too.

Welcome to As the Web Services Turn.

Timothy Appnel at O’Reilly Web Services DevCenter ponders the current (poor) state of RSS feed quality.  He covers some basics about RSS as well as things that can be done to improve its effectiveness.

This is classic.  I tried installing OpenMosix on two machines the other day.  I read the docs so I could have done it by hand, but I was lazy, so I .RPM’d it.  I rebooted, selected the OpenMosix kernel.  It did not boot.  I was a little confused because the bootloader seemed to think that my hard drive was ext2fs, but in both cases it’s ext3fs.  The new release solves this problem.  From the change notes:

The most important bug that was fixed in this release is the ext3/initrd problem, which caused machines with ext3 filesystems to not boot. This release compiles and runs with gcc-3.2. The value of HZ was reverted to 100 until the userspace utilities get modified. Makefile changes include openMosix version reporting and minor code cleanups. The /proc interfaces now work with fast machines. Bugs that are still present include a locking problem which occurs when using SMP and /dev/shm, and the reported problem with ln and /mfs.

How cool is that, my problems are solved.  I think it’s time to snag some new .RPMs and see what happens.

ONLamp: The SSH Crypotsystem.  A good intro for newbies. [via OSDir]

I’m in the middle of catching up on overnight news, and I have one news item reported in two completely different ways.  There was a fire at the network operations center at the University of Twente in the Netherlands.  From The Register:

With the fire “about the largest stock of hardcore pornography, illegally downloaded CD’s, movies and hacked / cracked software (applications and games) will also be turned into, er, ‘digital ashes’”, he notes.

Now Slashdot also covered the fire, choosing to focus on the losses suffered by the Debian development team.  UT happend to host several machines for the Debian project.  Here’s a snippet of their coverage:

Around 08.00 CET today the University of Twente Network Operations Center, which amongst other things hosts a SURFnet PoP as well as security.debian.org and non-us.debian.org, caught fire.

Here is the closing paragraph from The Register’s coverage:

This perceived misuse of Internet resources caused former Dutch education minister Loek Hermans to comment: “It would be nice if the students at Twente University would use their fast connections for information and education purposes, instead of downloading huge amounts of porn.”

I think that this was a bit of bad reporting, a chance to take a pot shot at p2p file trading and the RIAA.  A lot was lost in this fire.  Computers, infrastructure, buildings, and The Register is focused on the loss of illegal porn and music.  I’m normally amused by The Register’s tounge-in-cheek coverage, but I don’t think I was amused this morning.  Allow me to close.  I don’t know if you see it in that last quote or not.  He’s wishing that TU’s internet connection could be used for education and information, and Debian is mourning the loss of several boxen hosted at TU.

I wish I could smack the guy and say, “IT WAS!  IT WAS!”

I digress.  Here’s some coverage of the Debian servers involved (from Wichert Akkerman):

At around 8 this morning (local time) a fire started in the
computing facilities of Twente University. This affects Debian, since
on of our servers (satie) is hosted there. At this moment it seems
very likely that the machine can not be recovered from the fire.

The following services are currently down as a result of this:

  security.debian.org
  non-us.debian.org
  nm.debian.org
  qa.debian.org

We are working to restoring these services on another machine and
hope to have things in mostly working order by tomorrow. Security
advisories are still available at http://www.debian.org/security/

Mark Pilgrim:

By looking at the sites I currently read in my homegrown news aggregator, looking them up in the blogging ecosystem and seeing which sites they link to, weighing them slightly by popularity (based on the natural log of their incoming links) but also dividing by the number of other sites they link to (because a midlist site that only links to a handful of people is more relevant than a popular site that links to 100), and filtering out sites that don’t have RSS feeds (plus a few I already know I don’t want to read), I came up with a list of recommended new reading.

You can point his project, NewDoor, to your blog or links page.  His code is GPL’d, and can be found here: newdoor-1.0.tgz.

Simon Fell:

Is WSDL too hard? In response to Greg Reinacker’s comments I didn’t say WSDL is hard, I said its cumbersome and unproductive. Come on, its just angle brackets how hard can it be ??

I was gonna quote Greg’s post directly, but Simon’s got it.

This is a case of some interesting role reversal.  I went ot a friend’s house tonight after work to install a new cd burner.  The burner installed quite easily (Dell’s new desktop tower case is extremely easy to work in), and soon we were off to install the old burner that hadn’t worked correctly at her work so that she could transfer data back and forth over CD’s.  We took the old burner (HP8200 series external USB) over to her work and it preceeded to cause all kinds of problems.  I got a little Win2k bootup BSOD, then uninstalled the drivers/software.  This, of course, also somehow killed the drivers for the NIC, which is a really bad thing because she uses a client/server POS/inventory system.  I did manage to get everything up and running again.

After everything was done, she asked if I wanted the malfunctioning burner.  She used that nasty “if you don’t take it I’m just going to throw it out” line.  Of course I took it with me.  I’m like the Humane Society of random computer parts.  I have more crap floating around here because I can’t bear to throw away a working (or semi-working) piece of equipment.  I took it home, and for personal amusement I powered the burner up and plugged it into my Via Eden motherboard (running Red Hat 8.0), and rebooted.  My thought was that perhaps, as an off chance, Anaconda would pick it up on the reboot and it might possibly work.

I was a little dissapointed when no new hardware was found during the reboot.  After it booted, I did throw in Disk 1 of the Red Hat 8.0 distro just to amuse myself.  Amazingly, the CD-ROM automounted, a window popped up with the contents of the CD, and I was asked if I wanted to run the Autorun script or not.  How freaky is that?  It just worked.  Isn’t that how things are supposed to work under windows?

Linux on the desktop +1

Update: Oh yeah, it burns, too.

I hear from Freshmeat that GCC 3.2.1 is out.  There are a few bugfixes in the release, most significantly to IA-32 and x86-64.  You can look at a list of changes, release criteria, or you can snag it from your local mirror.

Jeremy Zawodny’s coworker Michael Radwin has some great coverage of ApacheCon 2002.

Anthill is the logical progression of the Ant build tool, and could be considered similar to Gump for a single project.  It makes automated building, testing, and deployment, and makes the stats easy to visualize.  Here are the changes to this version:

A build status indicator has been added to the main page. Now you can see whether the last build of any project succeeded or failed right on the main Anthill page. Tagging the source repository can be configured to happen always, only on successful builds, or never. The build email messages can now be customized based on a template. JVM properties can be configured for when Anthill invokes Ant. For example, the initial and max heap sizes can be set. Ant 1.5.1 is now included in the distribution of Anthill. There are also many bugfixes.