Mailing-List: contact dev-help@httpd.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@httpd.apache.org
Subject: Re: [PATCH]: Update request in connection based filters
From: Bojan Smojver <bojan@rexursive.com>
To: Apache Dev List <dev@httpd.apache.org>
In-Reply-To: <Pine.LNX.4.21.0209172321060.32729-100000@shell.ntrnet.net>
References: <Pine.LNX.4.21.0209172321060.32729-100000@shell.ntrnet.net>
Content-Type: text/plain
Content-Transfer-Encoding: 7bit
Date: 18 Sep 2002 23:41:40 +1000
Message-Id: <1032356500.4424.171.camel@beast.rexursive.com>
Mime-Version: 1.0

There is another possibility to fix this, and that's within the module
itself (I've tested it and it works):

--------------------------------------------------
static void logio_create_req(request_rec *r) {
    ap_filter_t *f;
    conn_rec *c = r->connection;

    /* Tie connection filters with request */
    for (f = c->input_filters; f; f = f->next)
        f->r = r;

    for (f = c->output_filters; f; f = f->next)
        f->r = r;
}

static void register_hooks(apr_pool_t *p) {
    ap_hook_pre_connection(logio_pre_conn, NULL, NULL, APR_HOOK_MIDDLE);
    ap_hook_create_request(logio_create_req, NULL, NULL,
APR_HOOK_MIDDLE);
    
    ap_register_input_filter(logio_filter_name, logio_in_filter, NULL,
                             AP_FTYPE_CONNECTION);
    ap_register_output_filter(logio_filter_name, logio_out_filter, NULL,
                              AP_FTYPE_CONNECTION);
}
--------------------------------------------------

Since this looks like modules own business, it seems like a good
compromise to me. What do you guys think?

Also, I've think that the correct thing to do in mod_logio is to be
AP_FTYPE_CONNECTION + 6, since SSL is AP_FTYPE_CONNECTION + 5, and we
want to be before it in input filtering and after it in output
filtering. I have the module working with r->notes for now.

Bojan

On Wed, 2002-09-18 at 13:22, rbb@apache.org wrote:
> 
> You don't want to do this.  A connection based filter is connection
> oriented.  By definition, it has no concept of a request.  While this
> might make sense for HTTP, other protocol modules will have a much harder
> time with this change.
> 
> Ryan