On Wed, Apr 14, 2010 at 10:18 AM, Nicolas Bock <
address@hidden> wrote:
> ok, maybe we don't need it anymore. I'll take a look.
>
> nick
>
> On Wed, Apr 14, 2010 at 09:06, Jeff Hammond <
address@hidden> wrote:
>>
>> Nick,
>>
>> I commented out that test and saw no adverse affects.
>>
>> Jeff
>>
>> On Wed, Apr 14, 2010 at 10:01 AM, Nicolas Bock <
address@hidden> wrote:
>> > Hi Jeff,
>> >
>> > this is really a hack that was supposed to address something else. I
>> > agree
>> > that it's not really necessary and should be removed.
>> >
>> > We do have some logic in
configure.ac that should address the fixed
>> > versus
>> > free command line argument issue, but we don't have a case for xlf yet.
>> > I
>> > will see whether we can simply remove the test whether FC == F77 or if
>> > not
>> > add a case to the command line argument check.
>> >
>> > nick
>> >
>> >
>> > On Tue, 2010-04-13 at 14:33 -0500, Jeff Hammond wrote:
>> >
>> > I don't understand the configure test that returns this error:
>> > configure: error: F77 (mpixlf77_r) and F90 (mpixlf95_r) compilers are
>> > different. This is not going to work.
>> >
>> > Unlike other Fortran compilers, XLF is fairly strict about fixed
>> > versus free source form and other F77-F95 syntax differences, hence
>> > distinguishing these two compilers is important. They both have the
>> > same xlf backend but one instruments -qfixed and the other does not.
>> > Hence the objects generated with them will be compatible.
>> >
>> > I can manually hack around this but I'd rather have a workaround in
>> > configure such as --ignore-f77-f95-mismatch.
>> >
>> > Thanks,
>> >
>> > Jeff
>> >
>> >
>> >
>>
>>
>>
>> --
>> Jeff Hammond
>> Argonne Leadership Computing Facility
>>
address@hidden / (630) 252-5381
>>
http://www.linkedin.com/in/jeffhammond
>>
>>
>
>