Re: [VANILLA-LIST:2237] xsg stuff

[Kurt Siegl <007@freemail.at>]

On Fri, 02 Apr 1999 you wrote:
>On Fri, Apr 02, 1999 at 12:13:36AM -0700, Jonathan Ellis wrote:
>> 
>> out a frame.  My question is this: IIANM, HANDLE_SIG and SIGNAL are both
>> #defined to the system call signal().  (Any reason why?)
>
>HANDLE_SIG and SIGNAL are redundant definitions.

Not really. There are three types of signals.

1) BSD signals. Once set they are active until you explicitly unset.
It defines
	SIGNAL = signal
	HANDLE_SIG =

2) POSIX signals. Automatically resets the signal handler after the signal
is caught and handled. It requires the signal to be set again in the handler.
It defines
	SIGNAL = signal
	HANDLE_SIG = signal

3) SYSV signals, behave like POSIX but you have an additional sigset
command which behaves like the BSD signal command.
It defines
	SIGNAL = sigset
	HANDLE_SIG = 

>> Does the signal handler need to be re-specified after each interrupt,
>
>The handler needs to be reinstalled after the signal is caught and handled.

Only in case of pure POSIX signals. That's what the HANDLE_SIG
define is good for. Linux and Solaris have pure POSIX signals.

All three types are correctly autodetected.

Kurt (007)

--
Kurt Siegl / Franzberg 4, A-4483 Hargelsberg, Austria
Email: 007@netrek.org       Tel (ISDN):   *(7225)7017
URL:   http://www.ooenet.at/user/siegl/kurt/