httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ralf S. Engelschall" <...@engelschall.com>
Subject Re: cvs commit: apache-1.3 STATUS
Date Sun, 16 Aug 1998 15:24:25 GMT

In article <199808161450.KAA21994@devsys.jaguNET.com> you wrote:
> Jim Jagielski wrote:
>> 
>> Ralf S. Engelschall wrote:
>> > 
>> > 
>> > In article <19980816040009.10720.qmail@hyperreal.org> you wrote:
>> > > marc        98/08/15 21:00:09
>> > 
>> > >   +Unix 1.3.2 RELEASE SHOWSTOPPERS:
>> > >   +
>> > >   +    * header detection is broken; eg. sys/resource.h often requires

>> > >   +      other includes first, etc.  This breaks things like RLimit*
>> > >   +      on many platforms.
>> > >   +
>> > 
>> > Do I understand it correctly? The _detection_ itself is broken, only.  Because
>> > when ap_config_auto.h then exists, these are only defines and not includes.
>> > The problem is that when we test for sys/resource.h we have to include another
>> > header in TestCompile, right?  Can we solve it be adding a list of
>> > dependencies to Configure?  Has anyone already general dependecies at hand?
>> 
>> I'm guessing that TestCompile would need to "include" some sort of
>> temp file that maintains the current header files being included..

> Hmmm. I'm think that if we:

>   1. Change the ordering of ap_config_auto.h so that the -CFLAG->#defines
>      are before the header files and
>   2. The TestCompile dummy program that checks header files does
>      an include of ap_config_auto.h as well as ap_config.h

> that this might work. Not sure if #1 will break anything since when
> building Apache the CFLAG->defines are ignored (since they are defined
> anyway). In fact, it might be better since it forces that included
> header to have the same #define values as when Apache is being built :/

> Think this might work, or is worth a try?

Sounds reasonable as a first step of improvement. The CFLAGS->#define are
harmless because they are surrounded with #ifndef..#endif. But I'm not sure if
this solves the problem of having to #include some other headers for some
headers we test. Or do you think they are then automatically included because
of ap_config.h? 

Marc, have we a failure-example at hand? I mean a platform where we need
sys/<whatever>.h for sys/resource.h so I can compare what would happen when we
include ap_config.h?
                                       Ralf S. Engelschall
                                       rse@engelschall.com
                                       www.engelschall.com

Mime
View raw message