[109950] trunk/dports/python/py-pyqwt/Portfile

Joshua Root jmr at macports.org
Fri Aug 23 06:49:14 PDT 2013


On 2013-8-23 23:10 , Ryan Schmidt wrote:
> 
> On Aug 22, 2013, at 14:19, michaelld at macports.org wrote:
> 
>> Revision: 109950
>>          https://trac.macports.org/changeset/109950
>> Author:   michaelld at macports.org
>> Date:     2013-08-22 12:19:12 -0700 (Thu, 22 Aug 2013)
>> Log Message:
>> -----------
>> py*-pyqwt:
>> * fix to use the right compiler;
>> * (temporary?) fix with sip 4.15.0_0 which generates extra interface
>>  code for a hidden overloaded virtual function;
>> * correct -universal compiling when Python is installed with +universal;
> 
>> Modified: trunk/dports/python/py-pyqwt/Portfile
> 
>> +        # if not universal, try removing all arch flags from the
>> +        # Makefiles.  If Python is installed as +universal, these will
>> +        # do the trick; if Python is installed as -universal, then
>> +        # these will do no harm.
>> +
>> +        if {![variant_isset universal]} {
>> +            reinplace "s at -arch \[^ \]*@@g" ${worksrcpath}/qwt5qt4/Makefile
>> +            reinplace "s at -arch \[^ \]*@@g" ${worksrcpath}/iqt5qt4/Makefile
>> +        }
> 
> So when not building universal, universal -arch flags are still ending up in these Makefiles. Where are these flags coming from? They're not supposed to be coming from python anymore now that this bug was fixed:
> 
> https://trac.macports.org/ticket/39669

And does it really do no harm to remove all -arch flags from the
Makefiles? How does build_arch get enforced?

- Josh


More information about the macports-dev mailing list