Rob McMillin wrote:
RM> open2? Has anyone tested this with spamd? The reason I ask is that I
RM> recently wrote a networking daemon in Perl that used open3() while doing
RM> the canonical close() on STDOUT, STDIN, and STDERR. I couldn't get it to
RM> work until I redirected those file handles to /dev/null rather than
RM> closing them. Probably something to do with the way open3() handles
RM> close-on-exec (FD_CLOEXEC) for filehandles other than those. I don't
RM> know if the current code handles this correctly or even does the
RM> close-on-exec, but it's a potential gotcha.
Seems to work fine for me. I've switched the require IPC::Open2 to use
IPC::Open2 but moved it to the top of the file. Also removed the = {} from the
hash declaration. All seems to work OK on my linux machine. Have been
horrendously unsuccesful getting DCC to compile on OSX (actually got most of the
way there on some of the progs, but not all of them). all tests seem to work OK
(except for the -S flag which I think has actually been broken for a little bit
now -- I'll investigate that tomorrow).
C
_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
