August 27, 2004

Annoying Proprietary Software

Mostly, I use free software. Not just because I'm a cheapskate -- this would be a poor reason since, as a student, most of my software is given to me, anyway. I like that it works well, that it has a tendancy to interoperate well, and that I can see its innards and twiddle them as needed.

There are, however, a few proprietary packages that my work requires that I use. To varying degrees, this is annoying the crap out of me.

To wit, and in rough order of annoyance:

IDL
A "rich man's" mathematical data analysis language, for people who like the idea of Matlab but were really more comfortable with Fortran back in the 70s.

Gigantic function library, but I can do most anything it can, often faster, and usually with superior (in terms of control and flexibility) results, with some combination of Perl, PDL, and graphics kits like XMGrace or Gnuplot. Since I'm too lazy to grab links for all those, feel free to Google them.

Plus the fact that half the time when I start it up, I can't actually use it, because the license server says that too many people are using it already, really drives home the un-freedom of the whole thing. I shudder to think how much of our group's experience is tied up in IDL code.

GRASP8-SE

Free student version of an extremely powerful physical optics package. The full thing costs $40,000 and change, so I don't see us springing for it in the immediate future.

Trouble is, the free version has some distinct limitations designed to induce you to open the wallet (or more likely, purchase order folder). For instance, models are artificially limited to two reflecting/scattering surfaces. Also, it only generally helps with our work, since it doesn't do refraction at all. Since our telescopes have lenses, this makes us somewhat uncomfortable. On a totally unrelated note, its data formats really show the Fortran under the hood, too.

I think it should be possible to implement the physical optics engine using PDL, so I printed out (and bound -- yes, I'm a freak, see below) the technical manual to learn how. If I have a really slow weekend in Israel, I may give it a try.

CODE V

Ray optics package.

I have mostly praise for this thing, actually. It works quite well, once you get to know it, has powerful features arranged in a way that suggests a modicum of thought may have been involved in their planning, and provides an amazingly complete palette of functionality for designing and analyzing an optical system.

The drawback. Only runs on Windows. I kid you not. Used to be a UNIX version, but they killed it for reasons that baffle me, considering that nearly every scientific shop on the planet is mostly *nix workstations. So I'm stuck either remotely logging into a Microsoft box, or, as I will be in Israel, running one in emulation inside my own system. The cryptographic hardware key is just icing on the cake, there.

But the fact that I'm not actively thinking about how to reimplement this one using free technologies is evidence of just how good it is, otherwise.

Posted by mill1974 at August 27, 2004 7:41 PM