Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!seismo!ut-sally!im4u!rutgers!princeton!rocksvax!martyl
From: martyl@rocksvax.UUCP (Marty Leisner)
Newsgroups: comp.sys.ibm.pc
Subject: Re: SIGINT question
Message-ID: <28@rocksvax.UUCP>
Date: Tue, 8-Sep-87 09:52:52 EDT
Article-I.D.: rocksvax.28
Posted: Tue Sep  8 09:52:52 1987
Date-Received: Wed, 9-Sep-87 05:58:23 EDT
References: <1930@isis.UUCP>
Reply-To: martyl@rocksvax.UUCP (Marty Leisner)
Organization: Xerox: Henrietta, NY
Lines: 33

In article <1930@isis.UUCP> ross@isis.UUCP (Ross McConnell) writes:
>
> I am using signal() in a MSC4.0 program, to catch CTRL-C - everything
>works fine. However, parts of the program consist of lengthy math
>routines, and the response to the CTRL-C can be quite slow (5-10
>seconds). I believe that this occurs because DOS only checks for
>CTRL-C while making BIOS calls. Is this true? If so, what is a good
>way around it. Make "dummy" BIOS calls in the math routines? Any
>suggestions would be appreciated.


I believe it will only catch control C on disk io unless you have break
on.  

I wrote a com port driver which I use for my console.  If the character 
received is a cntl-C, I immediately issue a call to interrupt 1BH (the
BIOS cntl-c handler).  It appears to work a little better than when I 
run on the console with the standard supplied IBM stuff
(it seems cntl-Cs are only checked as the DOS fetches
characters from the type-ahead queue).  If the program isn't catching
CNTL-Cs, it has the funny side effect of just printing  cntl-C on the screen --
at least I can tell the system is alive.

A reasonable way to handle the problem may be to write a TSR to look for
CNTL-Cs and if the program currently isn't in DOS, immediately do a far
call to the CNTL-C handler defined in the programs PSP.

I'm also looking for a way to abort a currently running program -- the 
above strategy should work for a special "hot key" which "aborts the current
process".  Any ideas on this?  Anyone have this toy?  (I'm tired of rebooting
when debugging software).

marty