Seema Alevoor wrote:
> On 08/14/08 20:51, Jeff Trawick wrote:
>   
>> Seema Alevoor wrote:
>>     
>>> I have updated the webrev ( http://cr.opensolaris.org/~seema/6670736/ ).
>>>
>>> Main changes include patches to apr_hints.m4 and apr_common.m4
>>> apr_hints.m4 - to add the threaded flags.
>>>       
>> Attached is an updated patch to solve:
>> * We can't change default for --enable-threads
>>     
>
> Hmm...in the same file, at couple of other places (for different platforms),
> the value of enable-threads was changed as below:
>       APR_SETIFNULL(enable_threads, [no])
>
>   
That's only for old FreeBSD and another obscure platform.  Defaulting to 
no here is only needed when autodetection would normally find a thread 
implementation (such as pthread.h + libpthread) but it is too deficient 
to use for APR.

>> * (Existing issue) Don't mess with large-file settings, because:
>> ** Adding ?-D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64? isn't necessary. 
>> APR >= 1.something already does ?-D_LARGEFILE64_SOURCE? by default.
>>     
>
> Defining -D_FILE_OFFSET_BITS=64 causes a failure when conifure tries to 
> enable -D_LARGEFILE64_SOURCE !
> ----
> configure: checking whether to enable -D_LARGEFILE64_SOURCE
> configure: result: no
> ----
> This is because off_t value differs when _FILE_OFFSET_BITS=64 is defined and 
> this causes
> the following condition to fail during configure.
>
>       if (sizeof(off64_t) != 8 || sizeof(off_t) != 4) {
>            exit(1);
>          }
>
>   
Let me back up a little bit:
I didn't think we should define either of these largefile-related 
definitions because:
* The APR solution provides the required feature for our 32-bit builds.  
(I don't think we need any patch.)
* We don't want to risk incompatibilities with other builds of APR.

Were these large file settings added to
* enable large file support in APR (needed with an Apache 2.0 build; not 
needed with Apache 2.2 build)
or
* fix an application or other issue in the way that APR enables large 
file support

Thanks!


Reply via email to