You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Marc Slemko <ma...@worldgate.com> on 1997/07/09 23:59:33 UTC
Dean's fix for Solaris HUP problem
Why couldn't we always do the bind() before the dup() (erm... F_DUPFD I
mean)?
Oh yes, someone should check things like UnixWare and Intel SVR4 to see if
they have the same problem, as suggested by the person at Sun; because of
their similar socket implementation, they may have the same problem.
Re: Dean's fix for Solaris HUP problem
Posted by Marc Slemko <ma...@worldgate.com>.
On Wed, 9 Jul 1997, Dean Gaudet wrote:
> I was trying to go for the minimal impact fix... only impacting solaris.
I am simply wondering if it isn't easier for everything to just make that
change for all OSes and end the hassles, especially if some other Unixes
may have the same problem.
I sure as hell hope we don't run into any OSes which loose flags when
doing a dup() on a bound socket...
> Note that I didn't commit those doc changes into 1_2_X yet so it's not up
> on the site. If we're happy with this fix as a stopgap measure then the
> docs should be merged.
>
> Dean
>
> On Wed, 9 Jul 1997, Marc Slemko wrote:
>
> > Why couldn't we always do the bind() before the dup() (erm... F_DUPFD I
> > mean)?
> >
> > Oh yes, someone should check things like UnixWare and Intel SVR4 to see if
> > they have the same problem, as suggested by the person at Sun; because of
> > their similar socket implementation, they may have the same problem.
> >
> >
> >
>
Re: Dean's fix for Solaris HUP problem
Posted by Dean Gaudet <dg...@arctic.org>.
I was trying to go for the minimal impact fix... only impacting solaris.
Note that I didn't commit those doc changes into 1_2_X yet so it's not up
on the site. If we're happy with this fix as a stopgap measure then the
docs should be merged.
Dean
On Wed, 9 Jul 1997, Marc Slemko wrote:
> Why couldn't we always do the bind() before the dup() (erm... F_DUPFD I
> mean)?
>
> Oh yes, someone should check things like UnixWare and Intel SVR4 to see if
> they have the same problem, as suggested by the person at Sun; because of
> their similar socket implementation, they may have the same problem.
>
>
>