httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ben Laurie <>
Subject Re: Apology
Date Tue, 19 May 1998 18:01:08 GMT
Brian Behlendorf wrote:
> At 08:47 PM 5/18/98 +0100, you wrote:
> >It is regrettable that the interface to the child-spawner needs to
> >change, but the kludge of adding stuff to the request_rec was just not
> >acceptable (to me, at least - I'm willing to be overruled, though - but
> >not without some argument!).
> How about: by adding things to the request_rec you don't have to change the
> API in a way which breaks 3rd-party modules which use ap_call_exec?

Yeah, that is a possible argument, but if you poke around, you'll
discover that there are sometimes two different request_recs involved,
for example, from mod_cgi:

     * we spawn out of r->main if it's there so that we can avoid
     * waiting for free_proc_chain to cleanup in the middle of an
     * SSI request -djg
    if (!ap_spawn_child_err_buff(r->main ? r->main->pool : r->pool,
			         (void *) &cld,
			         &script_out, &script_in, &script_err)) {
	ap_log_error(APLOG_MARK, APLOG_ERR, r->server,
		    "couldn't spawn child process: %s", r->filename);

so that kludge is liable to lead to confusion and bugs. Well, it
certainly confused me, and I really didn't like the idea of passing a
request_rec and a pool to ap_spawn_child_err_buff(), or two
request_recs. And it wasn't at all clear to me how to get away with only
passing one.

Also, unless 3rd party modules use ap_spawn_child_err_buff(), they are
already broken.

> Did the MMN get bumped?

Nope, I forgot that bit again. Coming RSN.

>  Did docs/new_features_1_3.html#api get touched?
> Or docs/upgrading_to_1_3.html?

I didn't know they should be.



Ben Laurie            |Phone: +44 (181) 735 0686|  Apache Group member
Freelance Consultant  |Fax:   +44 (181) 735 0689|
and Technical Director|Email: |
A.L. Digital Ltd,     |Apache-SSL author
London, England.      |"Apache: TDG"

View raw message