Mailing-List: contact new-httpd-help@apache.org; run by ezmlm
Precedence: bulk
Reply-To: new-httpd@apache.org
Date: Sun, 27 Aug 2000 03:37:16 -0700
From: Greg Stein <gstein@lyra.org>
To: new-httpd@apache.org, trawick@ibm.net
Subject: Re: return value of a filter
Message-ID: <20000827033716.A14349@lyra.org>
Mail-Followup-To: new-httpd@apache.org, trawick@ibm.net
References: <200008261933.PAA17041@k5.localdomain>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
User-Agent: Mutt/1.2i
In-Reply-To: <200008261933.PAA17041@k5.localdomain>;
 from trawickj@bellsouth.net on Sat, Aug 26, 2000 at 03:33:12PM -0400

On Sat, Aug 26, 2000 at 03:33:12PM -0400, Jeff Trawick wrote:
>...
> I think a filter and thus ap_pass_brigade() should return apr_status_t
> and, if appropriate, return APR_SUCCESS (all input bytes processed) or
> an APR error code.  An additional filter parameter could be used if
> required to tell the caller how many bytes were processed|written.

+1 on using apr_status_t.

The number of bytes sent is (as you point out) rather bogus. "sent" is a
meaningless concept from the (theoretical) filter chain standpoint. The
"consumed" notion is nice, but the design point is simply: I consumed the
whole darn brigade. Byte counts are moot -- the caller is certainly not
going to do anything with them.

-0 on any kind of "out of band" query facility w.r.t bytes process/written.

Cheers,
-g

-- 
Greg Stein, http://www.lyra.org/