This is a test cause I came back from doing a DX workshop to find I'd lost a 4Gb system disk on my Mac which contains all my mail and addresses so I hope this gets through. I'm reluctantly using a PC til I get the real machine running again (theoretically it was Retrospect backed up).
That said: at the workshop we used 9-10 copies of the 4.0.10 SGI 6.5 build and a single copy of one of the Linux builds (I didn't do the install and never actually got the details on which OS flavor that was but I do believe it was also 4.0.10). There were a variety of smallish problems in the Linux version, which I'm hoping the sysadm/new DX user who did the install will report directly to this list. One I think is well-known: arrows on slider/stepper interactors work either once or never, but you can type text into the field and change values that way. This was not a problem on the SGI version, just for comparison, so is probably a Lesstif ort still being worked on. The big apparently missing item was that the html/pages subdir seems to have gone missing in the 4.0.10 tarball for SGI. Can someone check that assertion? We were able to read the online pages from opendx.org during the class, but this was a funny omission: the top level pages were there, i.e., there is an html dir but no pages under that. The Linux tarball apparently did have them. I didn't have a lot of time to investigate this, but oddly we had to set DXMEMORY to 128 on the 192mb O2s to get dx to start. Anything greater generated the shared segment error we all know and love. Any ideas why? We also hit that weird problem where red lines appear in Image, and in one case, contours draped over a 2.5d surface suddenly decided to become infinitely tall and shot out of the image though the surface was still correct. Can't remember the details but I don't think I ever saw this in 3.1.4b (this was on the SGIs). I believe it happened after I showed the class how to end-run Rubbersheet by Computing positions based on elevation. But for the most part, things worked very well. We didn't attempt to use the imagemagick stuff cause I still haven't quite figured it out, though we did save TIFFs OK and viewed them in imagemagick. There was interest in activities such as javadx might provide solutions for, but no time to investigate that in any detail. Interestingly, this workshop was something of an AVS-openDX shootout. The organizers originally wanted the students to go through 1 day of AVS and 1 day of DX. But they also were inviting beginners to bring their own data. I managed to get them to split the class into 2 so we could have at least 2 days on whichever package they chose. In fact, the DX class did quite well, many actually did bring data and got it working, though that class length is too short in my opinion for real long-lasting learning. Nevertheless, there seemed to be a general appreciation of DX's potential, in no small part due to the sharp folks who attended. I'm not sure if it was simply an oversight or if the organizers chose not to embarrass AVS by having a final show-and-tell, cause I had a number of people who had completed cool projects with their own data that it would have been nice to show to the assembly, but I didn't see much completed work on the AVS side, probably since there were only 1-2 students left at the end of the 2nd day, while I still had 10 of the 14 who started. (sorry, but I really was feeling pretty smug; I did not attribute their staying to my personal charm, but to the fact that DX is actually a lot of fun once you get over the first couple of speedbumps) Something very interesting about AVS was revealed there, amidst some misinformation being spread about openDX by parties more inclined to AVS (particularly the overheard assertion that IBM had abandoned the market cause AVS was beating them at it; I saw no point in challenging this person who was obviously too clueless to correct). One hard-core long time AVS presenter admitted that he had to use DX for a job requiring streaklines since there is no way to do that in AVS. He also said that his group found errors when running AVS on a 64-bit architecture, chased this all the way up the chain at AVS to the president himself who (recently) flatly stated that it was not strategic for AVS to port to 64-bit. That floored me: I have to wonder if AVS is in the death spiral where income can't support R&D ensuring a future for the product. How can you not do a 64-bit port when all chips are going that way? Of course at the moment, DX is not 64-bit either but some of us know that it not only can be done but has been done for one large and influential client, and I believe at least some of you are working on a "public" version of same, correct? So keep it up! (Maybe that large and influential client would like to post it??) And another story he told revealed much about the root causes of AVS. Now, first I must say that we actually had a Stellar and a very early copy of AVS, about 11 years ago. It was impressive for about 5 minutes, back then, to waggle 1000 atoms (sphere primitives) around on the screen, thanks to the custom Stellar graphics architecture; none of us had time to figure out the wacky programming environment, so we set one student programmer at it and a semester later he had written one small program that none of us could understand either; that's about when my interest in using AVS ended. But this speaker revealed that in fact the original AVS was written precisely for the Stellar sales guys as a cool demo of the hardware. That is, AVS >started< as a dog and pony show for trade shows! I never knew that. I find it hilarious. (For anyone reading this who is relatively new to DX, you should know that DX sprang from the minds of several people who wrote seminal papers on fiber bundles, a schema for describing mathematical mapping systems that appeared to be just the ticket for a meta-solution for encoding data and spatial coordinate relations across many data types: DX, the program, is simply a realization of this well-founded mathematical basis system: oh, as a second derivative of its original function, it does do dog and pony work as well (:-) ). Chris Pelkie Vice President/Visualization Producer Conceptual Reality Presentations, Incorporated 30 West Meadow Drive Ithaca, NY 14850 Vox: 607-257-8335 (Alternate Daytime: 607-254-8794) [EMAIL PROTECTED]
