"Hoehle, Joerg-Cyril" Joerg-Cyril.Hoehle@t-systems.com writes:
It appears that the code in CVS provides a different implementation of getpid than what I sent to the slime list a few weeks ago. I don't know why there was a change. I saw no mention of a problem in this mailing list since. The current CVS is broken in some aspects (portability, FFI declaration, missing interface). Did my code break something?
The CLISP/Win32 returned -1.
I think that my own code, in a win32-clisp (not cygwin) 2.33.1 or older, without the WIN32 package, will define getpid to the constant -1. If this causes a problem, please tell me and I can change that. Possibly processes will not be interruptible. clisp >2.33.2 should work. I appreciate it when new code still works with old versions of other programs.
Well, the idea behind getpid is that Emacs can send a SIGINT to that process to interrupt it. That ideas does apparently not work so well on Windows but it certainly will not work if getpid returns -1.
I guess SIGINT is not the Windows way and something else should be used instead.
Helmut.