httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Randy Terbush <ra...@Covalent.NET>
Subject Re: MMN stuff
Date Thu, 13 Aug 1998 02:52:24 GMT
This was not the intent. Fixed. Thanks for pointing this out.

Rasmus Lerdorf <> writes:
> Just tried to compile mod_php3 against this new MMN stuff:
> functions/apache.c: In function `php3_info_apache':
> functions/apache.c:122: `MODULE_MAGIC_NUMBER' undeclared (first use in
> this function)
> functions/apache.c:122: (Each undeclared identifier is reported only once
> functions/apache.c:122: for each function it appears in.)
> make: *** [functions/apache.o] Error 1
> Was it really the intent to not make this backward compatible?  This is
> extremely ugly.  I have MODULE_MAGIC_NUMBER checks all over the place to
> make sure mod_php3 works with all versions of 1.2.x and 1.3.x.  Now it
> looks like I have to check for the presence of src/include/ap_mmn.h, and
> if it is there do something like:
> Except, I can't #define one macro to the value of another like this, so I
> am a tad screwed.  So I do this:
>    aplog_error(NULL, 0, APLOG_ERR | APLOG_NOERRNO, php3_rqst->server, log_message);
> #else
> #if MODULE_MAGIC_NUMBER >= 19970831
>    aplog_error(NULL, 0, APLOG_ERR | APLOG_NOERRNO, php3_rqst->server, log_message);
> #endif
> #else
>    log_error(log_message, php3_rqst->server);
> #endif
> As if this code wasn't ugly enough already.  And ok, I realize that in
> this case, just the mere presence of ap_mmn.h will also mean that MMN is
> more recent, so I could remove a check, but moving forward that isn't
> necessarily true.
> All other modules will of course instantly break as soon as we release
> this version of Apache and all module authors will have to scramble to
> release new versions in order to have them compile against 1.3.2.  Just
> doesn't seem like a very friendly thing to do in a point release.
> Why not simply keep MODULE_MAGIC_NUMBER and just add
> -Rasmus

View raw message