Re: cvs commit: apache-2.0/src/lib/apr/test ab_apr.c htdigest.c

[Martin Kraemer <Ma...@mch.sni.de>]

On Fri, Dec 03, 1999 at 04:12:33PM -0000, rbb@hyperreal.org wrote:
>   Programs that build using APR no longer have access to the HAVE_FOO_H
>   defines.  This was causing errors in these two test programs.  Fixed now.
>   -#ifdef HAVE_STRING_H
>    #include <string.h>
>   -#endif
>   -#ifdef HAVE_STDIO_H
>    #include <stdio.h>
>   -#endif
>   -#ifdef HAVE_STDLIB_H
>    #include <stdlib.h>
>   -#endif
>   -#ifdef HAVE_SYS_SIGNAL_H
>    #include <sys/signal.h>
>   -#else
>    #include <signal.h>
>   -#endif

Ehm... do I misunderstand what's happening here? What do you expect
to happen on platforms without some of these headers, or when
it's invalid to both include <sys/signal.h> AND  <signal.h> ?

    Martin
-- 
<Ma...@MchP.Siemens.De>             |    Fujitsu Siemens
Fon: +49-89-636-46021, FAX: +49-89-636-41143 | 81730  Munich,  Germany

Re: cvs commit: apache-2.0/src/lib/apr/test ab_apr.c htdigest.c

[Ryan Bloom <rb...@raleigh.ibm.com>]

On Fri, 3 Dec 1999, Martin Kraemer wrote:

> >   -#ifdef HAVE_SYS_SIGNAL_H
> >    #include <sys/signal.h>
> >   -#else
> >    #include <signal.h>
> >   -#endif
> 
> Ehm... do I misunderstand what's happening here? What do you expect
> to happen on platforms without some of these headers, or when
> it's invalid to both include <sys/signal.h> AND  <signal.h> ?

I expect this to not work right now on those systems.  Programs that are
just using APR, shouldn't be including apr_config.h, that's what everybody
wants.  If that's what we want, then our test programs for APR should
follow the same rules.  Without apr_config.h, we don't have access to
HAVE_FOO_H.  When these test programs are moved into the Apache
directories they will have access to Apache's ap_config.h (or whatever
Manoj called it), and these checks can go back in.  Or, we can provide
other ifdef's for the time being.  Regardless, the HAVE_FOO_H definitions
have to go, or our test programs don't test what they need to test, namely
APR's build environment.  I guess we could provide a small autoconf
routine that just checks for the headers the test programs care about, but
that seems like over kill.

Ryan

_______________________________________________________________________
Ryan Bloom		rbb@raleigh.ibm.com
4205 S Miami Blvd	
RTP, NC 27709		It's a beautiful sight to see good dancers 
			doing simple steps.  It's a painful sight to
			see beginners doing complicated patterns.