Re: Signal weirdness

Re: Signal weirdness

[malte]

I just tried your example on Solaris 2.3 and found a different though still
wrong behaviour:

	fn sighup { echo 'caught sighup'; sleep 10 }

rc-1.5b1 + readline from bash-1.13.1

	The signal gets caught but the handler is not executed until
	<return> is pressed. The shell keeps running.

rc-1.5b1 without readline

	Signal gets caught, the handler executed but then the shell exits
	with exit code 1 and "read: Interrupted system call" message.

rc-1.4 without readline works fine.

Malte

Re: Signal weirdness

[Alan Watson]

My patched version of v1.4 without readline does NOT display this bug
under any of Ultrix v4.3, OSF/1 v1.3, or IRIX v5.1.1.1.

Alan.

Re: Signal weirdness

[Tom Culliton x2278]

I just did some more testing, and the signal problem occurs in:

- 1.5beta1
	with and without readline
	with and without SVSIGS defined

- patched versions of 1.4

BUT not in vanilla 1.4!  It looks like the question is, "what's changed
in the signal handling from 1.4 to 1.5beta1?"

Tom

Re: Signal weirdness

[Vincent Broman]

I get the symptom you describe with my rc1.4 + readline
on sunos4.1.3 .

Vincent Broman,  code 572 Bayside                        Phone: +1 619 553 1641
Naval Command Control and Ocean Surveillance Center, RDT&E Div.
San Diego, CA  92152-6147,  USA                          Email: broman@nosc.mil

Re: Signal weirdness

[Byron Rakitzis]

Aigh. I will take a look at this. Thanks for letting me know.

Signal weirdness

[Tom Culliton x2278]

I just got a nasty suprise with signal handlers.  While trying to test:

	fn sigwinch {eval `{/usr/bin/X11/resize}}

I discovered that rc (1.5beta1 with GNU readline linked in, under SunOs
(4.1.3) and SCO-ODT (1.1 and 3.0)) dies after running the handler for a
signal that isn't ignored.  For example setting the following:

	fn sighup {echo 'got sighup';sleep 10}

and then sending the shell a sighup (kill -1 11532) from another window
causes it to print the message, wait, and the shell to exit.  If the
signal handler ends with a return command you get an error message but
the shell doesn't die.

This is really ugly, anyone have a clue?  I'd test this without readline
and even dig through code if work wasn't so piled up on my desk.  Maybe
in a week or so.  In the mean time, can anyone else verify this?

Tom

Re: Signal weirdness

[Chris Siebenmann]

 When I try this on a rc 1.5 betadev-1 without readline, I get a
'read: Interrupted system call' error and rc exits.

	- cks