From: Paul
Date: Fri, 08 Jun 2001 00:09:12 +0100
Subject: Re: [TSSP] UNIX for Windows
Terry, All Terry wrote: > However, there is a wonderful -->free<-- UNIX shell > program at: > http://cygwin.com/ This is a terrific step forward, because it means that not only will the modeling programs run as normal, but also some of the supporting scripts will too. For example, I have a little script into which I pipe Terry's scope trace data. It looks to see what the time axis is doing, runs the model to the same range and steps, then FTs both sets of data and does the plots. The modeling software is entirely command line and input file driven specifically to be scriptable in this way. It soon becomes a pain running the commands manually, for example searching for the resonances. You run tforce, look at the output file, see what the input reactance is doing, choose another frequency, and so on. Tedious! Especially when you change the software and have to recompute 80 coils. So you write a short shell/awk script to do it for you. This is one of the keys to the power of unix. You have a vast toolkit of utility programs, and you string them together any way you want, using the scripting languages. This approach requires a change of head for Windows users. There, the trend is for large monolithic programs with lots of buttons. If there's a button for what you want to do, then fine, otherwise you're stuffed. With the toolkit and script approach, you're limited only by your imagination. For this reason I'd recommend getting to know the shell command language, along with the tools awk, grep, and sed. The windows binary compiled by gcc under cygwin will not run without the API library cygwin.dll, so the EXE files thus produced can only be used locally within the cygwin shell emulation. Not long ago, Marco modified v0.8 to run under windows, compiled with the Microsoft compiler to produce a set of standard windows binaries. Only minor changes were required, but perhaps Marco will give the next revision, 0.9d the same treatment. You'll then be able to run the modeling programs using Marco's binaries outside of the cygwin environment. You'll have to make your own arrangements for stringing the tools together, but Kurt has recently demonstrated one way to do this. I'll try to get another cut of the software put together in the next few days. It has a little bit more documentation now. Cheers, -- Paul Nicholson, Manchester, UK. --
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.