ZHAO Wei writes:
> How could I detect broken pipe ? And how could I do anything graceful on
> it ?
I assume you are talking about the SIGPIPE signal. looking at
scsh/doc/cheat.txt, I see there is a signal/pipe constant. The related
functions are in the cheat sheet after this. They led me to look at
scsh/doc/scsh-manual/syscalls.tex. There it says:
Note that scsh does not support signal handlers for ``synchronous''
Unix signals, such as signal/ill or signal/pipe. Synchronous
occurrences of these signals are better handled by raising a Scheme
exception. We recommend you avoid using signal handlers unless you
absolutely have to; Section \ref{sec:event-interf-interr} describes
a better interface to signals.
So it seems like the signal/pipe constant is meant for use to signal
other people, not for handling. sec:event-interf-interr is defined in
scsh/doc/scsh-manual/threads.tex, but doesn't seem immediately relevant
for this case. scsh/signals1.c seems to disable SIGPIPE. It's been a
while since I've dealt with SIGPIPE, but looking at the write(2) man
page, it seems if it's disabled you'll get an EPIPE instead, which in
scsh is errno/pipe. back to cheat.txt, I see
(with-errno-handler HANDLER-SPEC . BODY)
HANDLER-SPEC is of the form
((ERRNO PACKET) CLAUSE ...)
ERRNO and PACKET are variables bound to the errno error being raised.
There are two forms for handler clauses:
((ERRNO ...) . BODY)
(else . BODY)
ERRNO are expressions evaluating to errno integers.
which is also mentioned in the FAQ, but with no example. I see an
example in scsh/dot-locking.scm, so perhaps you just need to wrap your
body with a
(with-errno-handler
((errno packet)
((errno/pipe)
...error code here...
#f))
...regular path here...
then you'd be happy?
-bri
|