Re: cvs commit: httpd-2.0/server main.c

[Jeff Trawick <tr...@attglobal.net>]

jorton@apache.org wrote:

> jorton      2003/12/10 05:43:14
> 
>   Modified:    server   main.c
>   Log:
>   * server/main.c (suck_in_expat): Remove function, USE_EXPAT is never
>   defined.

Does anyone have clues about the other hacks at the end of main.c?

Isn't the #ifdef AP_USING_AUTOCONF hack useful only if apr/apr-util is 
statically linked with httpd?  (Not possible with current build system AFAIK.)

What about #ifndef SHARED_CORE_BOOTSTRAP?  I don't know about the symbol name, 
but the comment

/*
  * Force apr_password_validate() into the image so that modules like
  * mod_auth can use it even if they're dynamically loaded.
  */

makes it sound like it applies only to the old case where apr/apr-util is 
statically linked with httpd.

Re: cvs commit: httpd-2.0/server main.c

[Joe Orton <jo...@redhat.com>]

On Wed, Dec 10, 2003 at 09:24:38AM -0500, Jeff Trawick wrote:
> jorton@apache.org wrote:
> 
> >jorton      2003/12/10 05:43:14
> >
> >  Modified:    server   main.c
> >  Log:
> >  * server/main.c (suck_in_expat): Remove function, USE_EXPAT is never
> >  defined.
> 
> Does anyone have clues about the other hacks at the end of main.c?
> 
> Isn't the #ifdef AP_USING_AUTOCONF hack useful only if apr/apr-util is 
> statically linked with httpd?  (Not possible with current build system 
> AFAIK.)

suck_in_APR() is necessary to pull in all of libmain.la and libos.la
regardless of static/shared apr.  (--disable-shared still gives you apr
statically linked into httpd, yes)

> What about #ifndef SHARED_CORE_BOOTSTRAP?  I don't know about the symbol 
> name, but the comment

Yes, I think suck_in_apr_password_validate can too,
ap_hack_apr_password_validate should do the same job.