apr-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jeff Trawick <traw...@attglobal.net>
Subject Re: Altogether Broken OtherChild logic
Date Fri, 31 Jan 2003 15:29:29 GMT

William A. Rowe, Jr. wrote:

> I'm simply thinking of renaming it apr_proc_other_child_died() and 
> documenting
> it correctly.  Step two is determining how I can then identify and 
> report that
> case in the WinNT MPM, or automagically as a callback.


How about rename it to

apr_proc_other_child_some_random_process_died_please_see_if_you_give_a_shit()

That's more accurate.  The caller doesn't know if it was registered as 
"other child" or not.

I'd rather have apr_proc_other_child_check() take an optional apr_proc_t 
(or have flavor called apr_proc_other_child_check_one()) and have it see 
if it cares about that process and if so see if it is still alive. 
Seems more useful.  Yeah, an extra rare syscall for some Apache MPMs but 
I feel that the function as it is just doesn't have enough generic 
usefulness to keep around.

Or just have apr_proc_wait() take care of calling any necessary other 
child maintenance function itself (if it finds out some process died and 
it was registered as other child).



Mime
View raw message