[g2clib] Build with -fpic on 32 bit too. Use -fpic on supported platforms over -fPIC

[Peter Robinson]

On Tue, Dec 11, 2012 at 9:57 PM, Orion Poplawski <orion at cora.nwra.com> wrote:
> On 12/11/2012 02:41 PM, Peter Robinson wrote:
>>
>> commit 8949ba49ab66e8642a38087fd19217dabc4f117b
>> Author: Peter Robinson <pbrobinson at gmail.com>
>> Date:   Tue Dec 11 21:40:52 2012 +0000
>>
>>      Build with -fpic on 32 bit too. Use -fpic on supported platforms over
>> -fPIC
>>
>>   g2clib.spec |   15 +++++++++++++--
>>   1 files changed, 13 insertions(+), 2 deletions(-)
>> ---
>> diff --git a/g2clib.spec b/g2clib.spec
>> index a00c138..607d095 100644
>> --- a/g2clib.spec
>> +++ b/g2clib.spec
>> @@ -1,6 +1,6 @@
>>   Name:           g2clib
>>   Version:        1.4.0
>> -Release:        1%{?dist}
>> +Release:        2%{?dist}
>>   Summary:        GRIB2 encoder/decoder and search/indexing routines in C
>>
>>   Group:          System Environment/Libraries
>> @@ -49,9 +49,17 @@ cp -p %{SOURCE1} .
>>
>>   %build
>>   CFLAGS="$RPM_OPT_FLAGS -DUSE_PNG -DUSE_JPEG2000"
>> -%ifarch x86_64 ia64 s390x ppc64
>> +
>> +%ifarch sparc64 s390x
>>   CFLAGS="$CFLAGS -D__64BIT__ -fPIC"
>>   %endif
>> +%ifarch x86_64 ia64 ppc64
>> +CFLAGS="$CFLAGS -D__64BIT__ -fpic"
>> +%endif
>> +%ifarch %{ix86} %{arm}
>> +CFLAGS="$CFLAGS -fpic"
>> +%endif
>> +
>>   make CFLAGS="$CFLAGS" CC="%{__cc}" ARFLAGS=
>>
>>
>> @@ -80,6 +88,9 @@ rm -rf $RPM_BUILD_ROOT
>>
>>
>>   %changelog
>> +* Tue Dec 11 2012 Peter Robinson <pbrobinson at fedoraproject.org> 1.4.0-2
>> +- Build with -fpic on 32 bit too. Use -fpic on supported platforms over
>> -fPIC
>> +
>
>
> So, is this the standard now?  -fpic over -fPIC and-fpic on 32-bit as well?
> I'm interested to learn more about this.

No, it's apparently faster on some architectures but if you're
enabling it on 64 bit you should on 32 bit as well.

Peter