CrayBay

Cray Inc. buys OctigaBay. OctigaBay was working on HPC-class computing equipment based on AMD Opteron processors and high-speed interconnect. The purchase is interesting, to say the least. The conspiracy theorist in me thinks that Uncle Sam doesn't want low-cost and "more commoditized" HPC gear generally available; the U.S. Government is Cray's biggest customer.

From OctigaBay's website:

Cray Acquires OctigaBay

Positions Cray to Address Entire High Performance Computing Market

On February 25, 2004, Cray Inc. announced a definitive agreement to acquire privately held OctigaBay Systems Corporation of Vancouver, British Columbia. OctigaBay is developing an innovative high performance computing (HPC) system designed to make supercomputing performance accessible to the growing community of scientific and technical computing users. This pending acquisition, coupled with Cray’s previously announced decision to commercialize the “Red Storm” system, will extend Cray’s product portfolio and multiply its addressable market by over four times.

Threadless

Despite the lack of any mention of cool t-shirts whatsoever, this thread about threads (and more) instigated, or rather, initiated by my friend Don Bailey on 9fans (or comp.os.plan9 for Usenet denizens) was part techrant, part flamewar, but more interesting that I expected. There were more than a few opportunities for it to degenerate into all-out name calling, but some decorum was maintained.

Remote Remote Control

Lately I've been running into remote control issues. That is, GUI desktop and server remote control technologies, particularly as espoused by VNC as well as others. A couple of questions occur:

1. Why are there so many VNC implementations? The code is GPL, I understand, but do we really need RealVNC, TightVNC, TridiaVNC, Ultr@VNC, etc.? I could understand perhaps the oddly named and separately maintained VNC client for some obscure platform, but most of these seem to be pretty much alike in terms of deployment, and differ on supported encodings and compression schemes/wire protocols. I guess I'm just hoping for more eventual cooperation and consolidation.
   

   
   

2. Why does RDP exist? I realize the answer is mostly "because Microsoft wanted it to", but it continues to stick around in Windows XP (and presumably will later as well) even though VNC is around. I guess the GPL is one reason. Besides, it's not like Microsoft has a history of just jumping on board with whatever open de facto standards are around, but it would be nice if RDP was more openly documented outside of what has been teased out of Microsoft's implementation and put together in rdesktop. It would also be cool if RDP could be just put into a multifunction VNC implementation.
   

   
   

3. Sun's Sun Ray? Yet another proprietary gizmo? Why? Again, at least some people are hacking it.
   

I did find an OS X native VNC server that looks okay, though I've not yet put it through it's paces, and have also managed to stick with an okay VNC client that doesn't crash, although it does lack some features and posses some bugs.

J and Scheme but not JScheme

An introduction to CS using J and Scheme side by side. Pretty neat.

Morbid Cognitive Dissonance

Joel on Lisp. I post this because it's a discussion about Lisp that some Lispers ought to read, although it's not a discussion that particularly interests me. For whatever it's worth, one of the points that hasn't been brought up in the discussion so far is that, as Graham says in the paper, with web apps users don't have to and shouldn't care what language the app is implemented in as long as it works. That means web app developers can work in whatever they like for whatever reasons and still provide a useful service without having to justify shipping something weird to the desktop OS.

Some of the points made in discussion regarding Lispers being jackasses with persecution complexes has a fair side and an unfair side to it. The fair side is that there are Lisp coders like that. The unfair side of is that it is, of course, a gross generalization. Nothing new here. People who really like something will defend it, and some of those people will become embittered to anything else. As for me, I just want to write in what I like and be left alone.

DocBook Dialogue

Following my quick-and-dirty post from an online chat with friends obstensibly about XML interop, Norman Walsh, co-creator and maintainer of DocBook sent me a kind email asking what I meant. I replied with the following (minor edits for the blog):

To be fair, I should clarify that that post, based on an IRC transcript, is misleading. I don't have a complaint about DocBook XML and interoperability with other XML formats per se. Instead, I have two complaints; one a broad criticism about XML interop as a whole, and the other a more detailed argument involving DocBook. I should further clarify that my criticism about DocBook isn't about the format itself -- I like DocBook a lot, and I want to use it, but I find it very hard to actually do so today, and most of this is due to a lack of good software.

I have a mildly more coherent, if more rambling (if such a thing is possible) rant about this. Briefly, there is and has been a lot of hype about how useful and easy it is to write documents in portable XM-based formats, DocBook being one example (and my chosen format), and I even knew some people who were doing it successfully, but it was nowhere near anything I considered easy, and certainly not as easy as the hype said it was. The post describes my particular scenario, which I'm sure doesn't hold for everyone, and keep in mind it *is* very much a rant, written with more than a smidgen of irritation at the time, but I would be thrilled for your comments on it and the issue at large.

For whatever it's worth, my interim solution has been to go back writing everything for work in ASCII plaintext first, and worrying about either document portability, be it semantic (as one might do with DocBook) or layout (Word, TeX) later. I've thought about doing DocBook again, this time eschewing XML-aware editors like XMetaL and the like, and just using a text editor, but still find the mess around XSLT and XSL-FO to be too daunting for too little return.

I asked Norm if I could blog our discussion, and he agreed, and posted his reply on his blog.

More to come soon. I'm catching up with everything that I missed during CodeCon, which was great.

They Lied To Us

[16:41] <@chris> ADTMag reports that the semantic web technologies are taking
        real form in the wake of recent W3C approvals and early pioneering work
        by vendors such as IBM, Boeing, Adobe and others."
[16:41] <@chris> ^^ snicker
[16:41] <@dnm> Man.
[16:41] <@dnm> It's like someone is playing a shell game.
[16:41] <@dnm> I mean, I'm still waiting for the much vaunted future promised
        in 1998 to arrive; that of XML interop.
[16:42] <@dnm> If you've ever tried to write a document in DocBook, you know
        what I mean.
[16:42] <@dnm> Let alone the Semantic Web.

Blog Bites

I've been meaning to post about a number of interesting things that I've either been pointed to or have found, but I haven't found time to do it. Instead, here's a quick list of some of them:

• The LLVM Compiler Infrastructure Project
  
• QNX Momentics, an Eclipse-based development suite hosted atop QNX Neutrino RTOS (now at version 6.2.1).
  
• 9grid, a Grid system based on a distributed Plan 9 network. See also Bell Labs' 9grid page. I hope to have a couple of my Plan 9 boxes on 9grid soon.
  
• Subversion. Nothing new here, but at breakfast a few weeks back, Ben Gross recently gave me enough first-hand anecdoatal evidence such that I installed it and tried it out, and I'm liking it a lot.
  
• Cooperative Linux looks really neat. Needless to say, I'm interested in how I and others might eventually be able to use similar technology and methods applied to old Lisp Machine operating systems.
  

There were more, but I'm forgetting them, and I can't track them down in my browser history. Shame on me for not bookmarking them. I've also inadvertantly toasted this post three times already, so I'm erroring on the side of caution and posting now. More to come later, if I can think of any more (and, of course, if I discover anything new that's cool!)