Serious KProcess Flaws

[Charles Samuels]

Ok, I'm sorry, this has been fixed in HEAD, I did a cvs up and nothing
happened, and then I did a cvs up -A and everything is nice again.

Apparently ossi is a wise person who fixed it, thank you!

On Thu, 26 Jun 2003, Charles Samuels wrote:

>
> I found an extremely serious bug in KProcess.  If one wants to use a waitpid
> themselves, they'll find that KProcessController randomly will waitpid(-1,
> ...).  This happens in KProcessController::delayedChildrenCleanup().
>
> I don't really understand the rationale for that function, so don't know how
> to fix it.  Please advise!
>
> Also, near the end of KProcessController::theSigCHLDHandler(int arg),
> waitpid(-1, ...) happens, on the condition that there is no
> KProcessController.  Why is KProcessController so arrogant as to think that
> it should 1) be the only SIGCHLD handler, and 2) insist that all children be
> started with KProcess?
>
> -Charles
>