[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: 20050818: HP/NAG Fortran NetCDF 3.6.0 Compiler Problems

"Von Engeln, Axel" <address@hidden> writes:

>> Well that's a bummer, isn't it?
>> 
>> Unfortunately I don't have the NAG compiler, but we have settings that
>> are reported to work:
>> 
>> http://my.unidata.ucar.edu/content/software/netcdf/docs/other-builds.html#Linux_nag-360
>> 
>> Let me know if this helps.
>
> Yes, it did help. Apparently the compiler needs -ieee=full settings as
> well. But I also had to include -dusty to make it work, thus the
> solution given above does not work. It stops in the testing with an
> error:
>
> Error: /var/tmp/test_write.020206.f: Argument 5 to NF_DEF_VAR is an
> array from TEST_NF_DEF_VAR, scalar from TEST_NF_REDEF
>
> Error: /var/tmp/test_write.020206.f: Argument 3 to NF_GET_VAR1_DOUBLE is
> an array from TEST_NF_SET_FILL, scalar from TEST_NF_REDEF
>
> -dusty degrades that to a warning but I am not sure if it is supposed to
> be this way. So maybe you want to look at the source code here.

The fortran 90 API and it's tests are due to come under close
examination any month now. New, more extensive, tests will be added.

>
> So, to round things up, HP compilation with a NAG compiler works the
> following way:
>
> CC=cc CPPFLAGS="-DNDEBUG -DNAGf90Fortran" CFLAGS="-O3" CXXFLAGS="" \
>    FC=f95 FFLAGS="-O3 -dusty  -DNAGf90Fortran -ieee=full" \
>    FLIBS='' configure --prefix=yourdirhere

Thanks, I've included that on the netCDF other-builds page:

http://my.unidata.ucar.edu/content/software/netcdf/docs/other-builds.html#HPUX_nag-360

Ed
-- 
Ed Hartnett  -- address@hidden