It is not a compilation problem, just that the ./configure do not want to
go to the ned. Eventually I did FC=NULL in environement variable and all
goes fine, netcdf and GMT were compiled and installed well. As I do not
use fortran it is not a big problem.
> Hi,
>
> Did you try the instructions for the PowerPC Mac? I know your MacBook
> has an Intel CPU, but it is possible the compiling will be the same as
> with the older Macs.
>
> Tim Hume
>
>
> On 20/9/06 12:03 AM, "Ed Hartnett" <ed@xxxxxxxxxxxxxxxx> wrote:
>
>> "Christophe Donnier" <donnier@xxxxxxxxxxxxx> writes:
>>
>>
>>>>> I can't ./configure netcdf 3.6.1 on a macbookpro (Mac os 10.4.7 )
>>>>> with X11. Seems that there is a problem with the GNU Fortran.
>>>>> If I set FC to NULL, it works, but with FC not set explicitly, or
>>>>> set to g77, I have the following error :
>>>>>
>>>>> configure: error: Could not link conftestf.o and conftest.o
>>>>>
>>>>> any idea of the problem ?
>>>>>
>>>>>
>>>>> thanks
>>>>>
>>>>> Christophe
>>
>>> Thanks everyone, GCC is version 4.0.1, here is a part of the
>>> config.log file, in fact I don't really need the fortran, but in case
>>> .....
>>>
>>> configure:5742: checking for Fortran preprocessor
>>> configure:5750: result: cc -E
>>> configure:5767: checking for C-equivalent to Fortran routine "SUB"
>>> configure:5774: gfortran -c conftest.f
>>> configure:5777: $? = 0
>>> configure:5789: result: sub_
>>> configure:5811: checking for Fortran "byte"
>>> configure:5819: gfortran -c conftest.f
>>> configure:5822: $? = 0
>>> configure:5824: result: yes
>>> configure:5841: checking for Fortran "integer*2"
>>> configure:5849: gfortran -c conftest.f
>>> configure:5852: $? = 0
>>> configure:5854: result: yes
>>> configure:5880: checking if Fortran "byte" is C "signed char"
>>> configure:5890: cc -c -Df2cFortran -g -O2 conftest.c
>>> conftest.c: In function 'sub_':
>>> conftest.c:4: warning: incompatible implicit declaration of built-in
>>> function 'exit'
>>> configure:5893: $? = 0
>>> configure:5896: gfortran -c conftestf.f
>>> configure:5899: $? = 0
>>> configure:5902: gfortran -o conftest conftestf.o conftest.o
>>> /usr/bin/ld: Undefined symbols:
>>> ___dso_handle
>>> collect2: ld returned 1 exit status
>>> configure:5905: $? = 1
>>> configure:5927: error: Could not link conftestf.o and conftest.o
>>>
>>>
>>> ============================================================================
>>> =
>>> To unsubscribe netcdfgroup, visit:
>>> http://www.unidata.ucar.edu/mailing-list-delete-form.html
>>> ============================================================================
>>> =
>>>
>>>
>>
>> If you don't need fortran, then by all means just skip it.
>>
>> Set your environment variable FC to null, and try again. (Do a "make
>> distclean" first).
>>
>> In sh, I can do it like this:
>>
>> make distclean && FC= ./configure && make check
>>
>> If you want the fortran interface, I would suggest getting the latest
>> version of gcc.
>>
>> There have been a lot of changes in gfortran in the last few versions
>> of gcc, so hopefully upgrading to the latest version of gfortran will
>> fix your problem, which I have never seen before!
>>
>> Thanks,
>>
>> Ed
>
> ==============================================================================
> To unsubscribe netcdfgroup, visit:
> http://www.unidata.ucar.edu/mailing-list-delete-form.html
> ==============================================================================
Christophe DONNIER
Responsable Informatique
*****************************
Collège de France
Chaire de géodynamique
Europole de l'Arbois
Bat Le Trocadero - Aile Sud
BP 80
13545 Aix en Provence - France
Email: donnier@xxxxxxxxxxxxx
Tel: +33 (0) 4 42 50 74 07
Fax:+33 (0) 4 42 50 74 01
Web: http://cdf.u-3mrs.fr
*****************************
==============================================================================
To unsubscribe netcdfgroup, visit:
http://www.unidata.ucar.edu/mailing-list-delete-form.html
==============================================================================
