I wrote:
>> > The new POSIX draft, at least, sanctions the automatic reset of
>> > SIGCHLD to SIG_DFL upon exec().

Terry Lambert appears to have written:
>> How does the "NOHUP" program continue to function in
>> light of this reset demand?

There is no ``demand'' involved.  The behavior of the system when
SIGCHLD is set to SIG_IGN has always been entirely
implementation-defined.  The new specification makes it clear that, in
the course of this implementation-defined behavior, implementations
are free to reset SIGCHLD to SIG_DFL when a new program is executed.
The new specification also makes it clear that the behavior of
functions such as system() and pclose() is undefined when SIGCHLD is
set to SIG_IGN.  (The semantics for SA_NOCLDWAIT are somewhat

This change was made because some implementors interpreted the base
documents as requiring that SIGCHLD be inherited across exec(), which
would then in turn require every program which ever makes use of
wait() to explicitly reset SIGCHLD (since the nominal default could
not be depended upon).

Giorgos Keramidas wrote:
> I'm not sure about POSIX, but the manpage of nohup does not mention
> SIGCHLD.  The only signals I see mentioned in revision 1.8 of nohup.1

That is correct.  SIGCHLD is entirely irrelevant to `nohup', as the
slightest amount of effort on Terry's part would have made clear.


To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message

Reply via email to