On 5/3/2017 5:25 PM, Brad King wrote:
> On 05/02/2017 05:47 PM, Christian Pfeiffer wrote:
>> Before filtering, gcc's output gives me the libgcc and libgcc_s twice.
> I think we already de-duplicate the results after extraction.
This only happens for the implicit library directories and frameworks
On 05/02/2017 05:47 PM, Christian Pfeiffer wrote:
> Before filtering, gcc's output gives me the libgcc and libgcc_s twice.
I think we already de-duplicate the results after extraction.
> However, I was mainly asking because the general matching logic there
> can break down in other ways, too.
On 5/2/2017 6:52 PM, Brad King wrote:
> On 05/02/2017 11:48 AM, Christian Pfeiffer wrote:
>> turn depends on libgcc_s. Any library starting with 'gcc' is being
>> ignored in CMakeParseImplicitLinkInfo.cmake L.136, so libgcc_s is being
>> stripped from the commandline. PGI Fortran however does not
On 05/02/2017 11:48 AM, Christian Pfeiffer wrote:
> turn depends on libgcc_s. Any library starting with 'gcc' is being
> ignored in CMakeParseImplicitLinkInfo.cmake L.136, so libgcc_s is being
> stripped from the commandline. PGI Fortran however does not need
> libgcc_s on its own and therefore