httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Marc Slemko <>
Subject Re: snprintf and release schedule
Date Mon, 20 Jan 1997 00:08:26 GMT
On Sun, 19 Jan 1997, Jim Jagielski wrote:

> Randy Terbush wrote:
> > 
> > The problem being that SO_LINGER does not work as advertised on
> > some systems. That was the whole point of lingering_close().
> > 
> The best solution, I think, would be to avoid SO_LINGER or any
> hacks that attempt to "emulate" it. It looks like a shutdown/close
> is supposed to be the answer on those SysV4 systems that require
> it. For other OSs, it's most probably safer to NOT do the
> shutdown, just in case.

I was thinking about this a bit and I don't see any reason (well... other
than what another brain-dead moronic bit of crap OS may do) why we
couldn't call shutdown() on all platforms.  

What that would do would be gain us would be less ifdefs and we would also
be able to tell more reliably if the data actually was delivered (ie. 
return code of shutdown()) and log an error if not.

Against doing that is that it would likely cause child processes to hang
around for longer while the shutdown() is going on.  Since we can't do
anything useful with any error we do get from the shutdown(), just doing
the close() and not caring if there is an error trying to deliver the
final data is probably just as good.

> We can wrap this around with something like "NEED_SHUTDOWN" or
> -- 
> ====================================================================
>       Jim Jagielski            |       jaguNET Access Services
>           |
>                   "Not the Craw... the CRAW!"

View raw message