[PATCH] child_init() is int rather than void..?

[Rodent of Unusual Size <co...@decus.org>]

    Erm, from http_config.h we have:

>    /* Regardless of the model the server uses for managing "units of
>     * execution", i.e. multi-process, multi-threaded, hybrids of those,
>     * there is the concept of a "heavy weight process".  That is, a
>     * process with its own memory space, file spaces, etc.  This method,
>     * child_init, is called once for each heavy-weight process before
>     * any requests are served.  Note that no provision is made yet for
>     * initialization per light-weight process (i.e. thread).  The
>     * parameters passed here are the same as those passed to the global
>     * init method above.
>     */
>    int (*child_init)(server_rec *, pool *);

    But the declaration of the "global init" method is:

>#ifdef ULTRIX_BRAIN_DEATH
>    void (*init)();
>    :
>#else
>    void (*init)(server_rec *, pool *);
>    :
>#endif

    and even child_init_modules() doesn't expect a return value.  I
    think this is new enough that changing it woudn't require bumping
    the MMN, eh?

    #ken    :-)}

Index: http_config.h
===================================================================
RCS file: /export/home/cvs/apache/src/http_config.h,v
retrieving revision 1.38
diff -c -r1.38 http_config.h
*** http_config.h	1997/07/19 20:27:53	1.38
--- http_config.h	1997/07/20 01:00:12
***************
*** 228,234 ****
       * parameters passed here are the same as those passed to the global
       * init method above.
       */
!     int (*child_init)(server_rec *, pool *);
  } module;
  
  /* Initializer for the first few module slots, which are only
--- 228,238 ----
       * parameters passed here are the same as those passed to the global
       * init method above.
       */
! #ifdef ULTRIX_BRAIN_DEATH
!     void (*child_init)();
! #else
!     void (*child_init)(server_rec *, pool *);
! #endif
  } module;
  
  /* Initializer for the first few module slots, which are only

Re: [PATCH] child_init() is int rather than void..?

[Dean Gaudet <dg...@arctic.org>]

+1

Dean