Mailing-List: contact new-httpd-help@apache.org; run by ezmlm
Precedence: bulk
Reply-To: new-httpd@apache.org
Date: Sun, 27 Aug 2000 05:19:38 -0700
From: Manoj Kasichainula <manoj@io.com>
To: new-httpd@apache.org
Subject: Re: [PATCH] ap_add_filter
Message-ID: <20000827051938.E3266@manojk.users.mindspring.com>
Mail-Followup-To: new-httpd@apache.org
References: <20000820180131.B32408@manojk.users.mindspring.com>
 <Pine.LNX.4.21.0008201826260.28794-100000@koj.rkbloom.net>
 <20000821223902.C9261@manojk.users.mindspring.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
User-Agent: Mutt/1.3.7i
In-Reply-To: <20000821223902.C9261@manojk.users.mindspring.com>;
 from manoj@io.com on Mon, Aug 21, 2000 at 10:39:02PM -0700

On Mon, Aug 21, 2000 at 10:39:02PM -0700, Manoj Kasichainula wrote:
> On Sun, Aug 20, 2000 at 06:29:11PM -0700, rbb@covalent.net wrote:
> > I would really like to see this design in code, because even with all of
> > the discussion about it, I don't actually see how some of the details
> > work.  In my mind, this design is more complex, but Manoj, you keep saying
> > this is simple.  Please provide some compilable code, and enlighten me.
> 
> It's coming (though I'd be pleased to see someone else beat me to it).
> I'd like to finish suexec first (which I don't think is far off).
> 
> It won't be interesting code, though, because I'll end up modifying
> one of the existing filters that doesn't really need to add filters.
> But the concept should still be clear.

OK, the filter chain management stuff doesn't look anything like I
expected it to. I see something that looks kind of funky to me, with
the module adding a hook so that it can add itself to the filter
chain, and also registering itself with the core seperately.

Why two seperate calls?

I'm still feeling my way through this stuff, but I probably would have
had each module return an ap_filter_t pointer in its filtering hook
instead, and let the core engine actually add the filter to the filter
chain. No coincedence that this "goes with" my filter chain management
proposal :)