Date: Fri, 2 Apr 1999 07:51:02 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
X-Sender: eliz AT is
To: "Mark E." <snowball3 AT usa DOT net>
cc: djgpp-workers AT delorie DOT com
Subject: Re: Bash ctrl-c problem simplified
In-Reply-To: <199903312046.UAA50148@out4.ibm.net>
Message-ID: <Pine.SUN.3.91.990402073955.22974D-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Reply-To: djgpp-workers AT delorie DOT com
X-Mailing-List: djgpp-workers AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk


On Wed, 31 Mar 1999, Mark E. wrote:

> When I compile and execute this program, I get a traceback instead of 
> what the documentation below (taken from signal) describes.

As far as I can see, the documentation only says the special behavior is 
when Ctrl-C is pressed.  In your case, you called `kill' instead of 
pressing Ctrl-C.

> Does the 
> special SIGINT behavior only apply when the ctrl-c key is pressed or 
> should it also apply to kill() generated interrupts?

I don't know.  This behavior didn't change in v2.02: as far as I can see, 
v2.01 also bypassed the traceback only if Ctrl-C was pressed.  We *do* 
want the traceback when SIGINT is generated by Ctrl-BREAK.  It is easy to 
change the library so that *only* Ctrl-BREAK causes the traceback, but 
I'm not sure this is the right thing to do.  For example, in the case of 
Bash, this would disable the traceback for Ctrl-BREAK as well, unless you 
do something special about Ctrl-BREAK inside Bash.

So I would like comments from people about this before we decide where 
this should be changed.

Incidentally, producing the Ctrl-C effect from a C function is very easy: 
just replace ``kill (getpid (), SIGINT);'' with the following snippet:

  __asm__ __volatile__ ("movb $0x79,%al;call ___djgpp_hw_exception");