Sriram Natarajan wrote:
> Jeff
> Thanks for such a nice review. Please see my response inline.
>
> - Sriram
>
> Jeff Trawick wrote:
>> Sriram Natarajan wrote:
>>> Hi
>>> We would like to link PHP MySQL extension with MySQL 5.1 runtime 
>>> instead of the current MySQL 5.0 runtime. Accordingly, here is the 
>>> patch that addresses this. Can you kindly let me know if this looks 
>>> ok to you
>>>
>>> http://cr.opensolaris.org/~sn123202/php52-mysql51/
>>
>> It looks like the MYSQLI_SHARED_LIBADD/PDO_MYSQL_SHARED_LIBADD/etc. 
>> respects settings made outside of configure. Can we make use of that 
>> to set the required MySQL runpath in Makefile.sfw instead by patching?
>>
> I had tried this during my very initial php integration days and for 
> some reason I had to patch our build process as passing it through 
> environment did not give me the intended results. Unfortunately, I am 
> not able to remember the exact reason today. I can investigate it 
> again - and provide you with more detailed analysis as to why I had to 
> patch the build process in the first place.  However, I would like to 
> address this - even if it turns out to be possible - post build 111 - 
> with a separate bug to minimize any possibility of introducing any new 
> regression issues for build 111a.
>
no clue here on why it wasn't respected; I grepped through configure and 
it seemed to only add to those flags

>> A common pattern for deciding which libmysqlclient to link in 
>> ext/FOO/config*m4 is
>>
>> if --enable-maintainer-zts; then
>> block-A
>> else
>> block-B
>> fi
>>
>> Would it be cleaner to patch the if statement to enable that leg of 
>> code, rather than replacing block-B with block-A?
>>
>> (something like this:
>>
>> if test "$enable_maintainer_zts" = "yes" -o 
>> "$use_reentrant_mysqlclient" = "yes"; then
>> ...
>>
>>
>> )
>>
> Makes sense. For this integration, I have modified this change and the 
> updated webrev is here
>
> http://cr.opensolaris.org/~sn123202/php529-mysql51.2/webrev/

cool; if you find how to pass the other envvars from outside configure, 
you can set the new variable use_reentrant_mysqlclient as well, to 
simplify the patch

btw, why is -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64 added to DEFS in 
php_Makefile.global.patch for these CRs?

Reply via email to