davem@bmers58.UUCP (Dave Mielke) (09/01/89)
HP-UX documentation says that SIGCLD should not be used as it may be removed in some future release. Is this really true? It is the only asynchronous way to detect the death of a child process and really is extremely useful. If it is going to be removed then when, and what are the suggested alternatives?
decot@hpisod2.HP.COM (Dave Decot) (09/02/89)
> HP-UX documentation says that SIGCLD should not be used as it may be > removed in some future release. Is this really true? It is the only > asynchronous way to detect the death of a child process and really is > extremely useful. If it is going to be removed then when, and what are > the suggested alternatives? I suppose they should use SIGCHLD instead, since that is required by POSIX, and SIGCLD isn't. Perhaps we should say that. Dave
rml@hpfcdc.HP.COM (Bob Lenk) (09/02/89)
The warning about SIGCLD originated in AT&T's System III manual. It is unlikely to go away in the near future. However, the POSIX standard defines the signal named SIGCHLD (following the Berkeley convention) instead. The best way to write code that will have a long, portable life is probably to use SIGCHLD according to POSIX; in particular this means not setting the action to SIG_IGN, but using SIG_DFL to ignore the signal. The name SIGCHLD is defined on series 300 releases beginning with 6.5, and I believe on all releases of the series 800. Bob Lenk rml@hpfcla.hp.com hplabs!hpfcla!rml