Ryan Bloom <[EMAIL PROTECTED]> writes:
> > > >why is SingleListen needed?
> > >
> > > Pretty much to make SINGLE_LISTEN runtime rather than compile
> > > time... Again, to give the admins more control over how
> > > Apache handles mutexing.
> >
> > I'm not totally against it, it just seems that it is for playing
> > around (sort of like USE_NONE_SERIALIZED_ACCEPT) or trying to work
> > around an Apache bug (i.e., maybe there is some platform/version where
> > we're supposed to turn on SINGLE_LISTEN_UNSERIALIZE_ACCEPT but don't).
> > Why can't they compile this in? This would seem to be useful in
> > extremely rare circumstances, and only for somebody who really knows
> > what they are doing.
>
> I disagree. Compiling in SINGLE_LISTEN_UNSERIALIZE_ACCEPT means that
> I have to no up front exactly how the server will be used. What if I am compiling
> one binary to be used on 50 machines (big fortune 2500 companies do this a lot,
> so that they have a common interface). On some of those machines (intranet),
> I may only want one port, so I want to be able to use S_L_U_A, on others
> (Internet), I want two ports (one SSL on regular), so I don't want to use S_L_U_A.
S_L_U_A already handles this capability. When S_L_U_A is defined, you
still get a mutex when there is more than one port. You skip the
mutex only when there is a single port (hence the leading "S_L_").
> I have always thought that being able to turn this on and off at run-time would
> be very cool and useful.
The directive would seem to only be useful for a platform/version that
has no thundering herd problem with accept() and where Apache's
ap_config.h has not yet been tweaked to indicate that. Is this a
real-life problem that we need to solve, or is it rare enough that the
user can use the existing compile-time mechanism until Apache's
ap_config.h is fixed and we can avoid adding another directive.
--
Jeff Trawick | [EMAIL PROTECTED] | PGP public key at web site:
http://www.geocities.com/SiliconValley/Park/9289/
Born in Roswell... married an alien...
