httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
Subject Re: RANT: Absolute Paths and configure
Date Tue, 04 Apr 2000 17:52:15 GMT
On Tue, 4 Apr 2000, Jim Jagielski wrote:

> wrote:
> > 
> > On Mon, 3 Apr 2000, Jim Jagielski wrote:
> > 
> > > My completely simple-minded approach is that APR should be configured
> > > first. It determines what capabilities we have. MPM should be
> > > configured after that and see which of those capabilities it
> > > needs to use. In this approach, I guess, MM needs to be configured
> > > before it all.
> > 
> > I don't believe MM needs to be configured before APR.  APR is currently
> > being configured before MM, and this seems to work just fine.  I don't
> > mind changing, as long as the APR configuration is ALWAYS independant.
> > 
> Can APR exist without MM? I thought that APR relied on MM for
> it's shared memory support. So it's up to MM to determine what
> flavor is available (mmap, shmget, whatever) and then APR uses
> that. I may be mistaken though about how APR uses MM though.

APR uses MM for it's shared memory logic, but the two programs are
completely separate.  Regardless of what MM decides, APR uses the same
function calls.  Take a look at it, and it should be obvious that APR can
be configured either before or after MM.

The only reason I bring this up, is because I would like APR's configure
script to be relatively simple.  If our configure script for Apache has to
call MM's configure script which then calls APR's script, and then it
calls Apache's configure, I think this is too complex.  Especially if it
is possible to just call APR's configure script and then Apache's script.

Again, my only concern is that APR is always configurable by itself (with
MM of course).  As long as this is done, anything else is good in my book.


Ryan Bloom               
406 29th St.
San Francisco, CA 94131

View raw message