You can read there for more details. But anyways, I've rewritten all my scripts on my new machine, and I've tried to make them as clean and modular as possible. In the open source spirit, I've made all of them available online for your amusement. Unfortunately for some, they're Unix only (this would be a good time to switch operating systems if you're using Windows!).
A quick rundown:
- There's scripts for playing on ICC, committing the current sources to CVS and tagging them (as well as updating a central version file), grabbing old sources, matching an arbitrary version against multiple programs in Xboard, and selecting a good version to test with the algorithm described in that old post.
- There's a sort of "header file" called ZINCLUDE. This has some basic definitions, where you can plug in your program's name, the directory where it's located, where other engines are, etc.
- Because there's not really an include mechanism for sh scripts, I hardcoded the path to ZINCLUDE in each script.
- Some other things will need modifying, like the CVS stuff in zget, and the stuff in ztests.c
- Allow testing on position suites, e.g. the Nunn positions
- Make it more modular, have a function in ZINCLUDE to calculate the next version number
- Make the C program more flexible, I just hardcoded the directories
- Make the list of other engines to test be an external file
You can browse the code here: http://zct.cvs.sourceforge.net/zct/Scripts/
...or you can check it out via anon CVS from SF.