ram@lscvax.UUCP (Ric Messier) (04/26/88)
Out of curiosity, we here are interested in knowing if anyone out there has implemented the NCAR graphics software and if so, on what kind of machine. We are currently attempting to get it up and running and would be interested in hearing from others, particularly to hear of problems they might have had. Thanks a lot! -- - Kilroy ram@lscvax.UUCP *** Can't deal, &CRASH 'Just what cowpatch is Lyndonville, Vermont in anyway?' Recursion - a procedure or function that calls itself SEE RECURSION
jim@pemrac.swri.edu (James Biard) (04/28/88)
In article <405@lscvax.UUCP> ram@lscvax.UUCP (Ric Messier) writes: >Out of curiosity, ... anyone out there has implemented the NCAR graphics ... Here at SwRI I have implemented the GKS version of the NCAR package on an HP9000 model 520 and on a Convex C1. This was using an early, crippled release of the package with the gut level GKS that NCAR provided. The NCAR people unreleased the version I had right after I got it, about 2 years ago (can it really have been that long?). Anyway, we are about to get the present release with a commercial GKS, and I will get to repeat the process. I can't wait to see what kinds of bugs are in the new release. (Not to flame NCAR really, they do good work there.) Soooo... here is one person that has installed the package. Jim Biard ----------------------------- jim@pemrac.UUCP "If we knew what we were doing, it wouldn't be research!" ---------------------------------------------------------
mcdonald@uxe.cso.uiuc.edu (04/29/88)
>Out of curiosity, we here are interested in knowing if anyone out there >has implemented the NCAR graphics software and if so, on what kind of >machine. We are currently attempting to get it up and running and would >be interested in hearing from others, particularly to hear of problems >they might have had. Thanks a lot! We have had NCAR on our VAX780 (VMS) for years. I didn't do the work of putting it on, so I don't know how much trouble it was. It works OK and is our normal graphics package. I have put various pieces of it on my IBM PC, including the 3d surface drawer and the countour plotter. I have not, however, tried to copy their device-independent run-time system. I'll give my impressions of the whole thing, though. First, most people around here actually like the thing. I don't myself, though. In the first place, the version we have contains some of the worst-written code I have ever seen. I have spent a lot of time and fuel feeding the flame wars in comp.lang.c defending goto's, but the stupid NCAR code is a perfect example of their abuse. It is truly awful. It would cause apoplexy in the goto- haters. But that's not all. The worst is yet to come. They allow one to call the same subroutine like line(x,y) with x and y being EITHER integer*4 or real *4, and DEPEND ON BEING ABLE TO TELL WHICH IT IS BY LOOKING AT THE BIT ENCODING PATTERN!!!!! Gasp! Choke Aargh! The run time system is a really clunky mess. It depends on lots of clunky intermediate output and a sluggish interpreter. The output to pen plotters is, usually, abysmally bad in terms of speed, because it picks up the pen at the beginning of every line segment, whether it needs to or not. The VAX version takes about 30 minutes to produce a typical 50x50 grid contour plot on our HP plotter, while my optimized (direct output to plotter, no intermediate file, and not picking up the pen unless needed) version takes 1 minute. On the other hand the high level plotting algorithms seem to be (relatively) OK. I'd like to know what machine you are trying to use it on. It is worth the price, though. Doug McDonald (mcdonald@uxe.cso.uiuc.edu)