Re: Print plot error

[Rob Mahurin]

On Aug 24, 2009, at 6:19 PM, Dmitri A. Sergatskov wrote:
> On Mon, Aug 24, 2009 at 4:39 PM, Rob Mahurin<address@hidden> wrote:
> > You probably want waitpid(-1,WNOHANG) so that print won't freeze  
> > if there is actually no child.
> >
> > Attached as a patch (which is not a changeset, sorry).  This would  
> > be good for 3.2.3.
>
> With WNOHANG I see few (one or two) defunct processes remains after
> loop of ~100 prints.

Sorry, I wasn't clear: it's safe to do
        while (0 < waitpid (-1, WNOHANG) ); endwhile

which cleans up any & all errant children, doing nothing if there are  
none.

That's not quite what I did in the patch, since I misunderstood  
waitpid's return values.

Ben Abbott asks:
> Also what about children that are not zombies? Will such produce an  
> infinite loop?

Not as written here (this one is tested).

> Regarding the "zombies" does waitpid() kill them or just wait until  
> the die?

No.  A "zombie" is a process that has already called exit() but is  
being kept in memory so that its parent can check its exit status.   
There's a little information in the man page for ps, and wikipedia  
has a "zombie process" entry.

Rob

--
Rob Mahurin
University of Manitoba, Dept. of Physics & Astronomy
at:     Oak Ridge National Laboratory   865 207 2594
        Oak Ridge, Tennessee            address@hidden

print.patch
Description: Binary data