Re: [VANILLA-LIST:2213] Problem with recent patch to update.c

[Dave Ahn <ahn@vec.wfubmc.edu>]

On Thu, Mar 25, 1999 at 06:23:28PM -0500, Dave Ahn wrote:
> 
> That's really funny.  SA_NOCLDWAIT is part of POSIX signals (as well as most
> modern SysV that support sigaction).  I thought Linux was supposed to be
> POSIX complaint.  Go figure.

Replying to my own post, I am mistaken.  Standards are so much fun.  POSIX
does not have SA_NOCLDWAIT.  It is a SVR4 thing.  After digging up some old
saved articles about cross platform signal compatibility, my memory has been
refreshed.

Ignoring SIGCHLD stops zombies if you use signal() rathern than sigaction()
and only with SysV signal() semantics.  This behavior doesn't seem to be
consistent anymore; it certainly seems to be broken under Solaris.  Most
modern OS's have SA_NOCLDWAIT flag to replace the old SysV signal() behavior.
At least, Solaris, IRIX, FreeBSD and all other SVR4 derivatives do.

If we want to support the old behavior, configure would need to check for
SA_NOCLDWAIT.

-- 
Dave Ahn <ahn@vec.wfubmc.edu>        |  "When you were born, you cried and the
                                     |  world rejoiced.  Try to live your life
Virtual Endoscopy Center             |  so that when you die, you will rejoice
Wake Forest Univ. School of Medicine |  and the world will cry."  -1/2 jj^2