[MITgcm-devel] g77 on darwin does not compile w2_print_e2setup.F

[Jean-Michel Campin]

Hi Martin,

I am going to verify that your latest suggestion works well,
and if it does, will check-in this.
Thanks,
Jean-Michel

On Wed, Jun 03, 2009 at 04:57:56PM +0200, Martin Losch wrote:
> Hi Jean-Michel,
> I did not mean to propose these changes, I just wanted to show you what 
> makes the code compile (it's clear that char*2 is not what you want).
> I had already tried your suggestion, same problem
> I also tried
> CHARACTER*2 cc1
> and then cc1(1:1), but that does not work either. After all it's a  
> compiler bug.
>
> This works (but will it give the results you expect?):
>>       CHARACTER*1 edge(0:4)
> [...]
>>       DATA edge / '?', 'N' , 'S' , 'E' , 'W' /
>
> [...]
>>             WRITE(W2_oUnit,'(2(3A,I3),A,4I3,A,2I6)')
>>      &      '  ', edge(i), '.Edge Facet', j, ' <-- ',
>>      &           edge(ii), '.Edge Facet', jt,
>
> [...]
>
> Martin
>
>
> BTW this is my g77 compiler:
>> sysm15::build> g77 -v
>> Reading specs from /usr/local/lib/gcc/i686-apple-darwin8.8.1/3.4.0/ 
>> specs
>> Configured with: ../gcc-3.4.0/configure --enable-languages=f77
>> Thread model: posix
>> gcc version 3.4.0
>
> On Jun 3, 2009, at 4:46 PM, Jean-Michel Campin wrote:
>
>> Hi Martin,
>>
>> Sorry, but with the changes you propose to make it to compile:
>>>>     CHARACTER*1 edge(4)
>>>>     CHARACTER*2 cc1
>> it's still not very clear why it is a problem in the 1rst place
>> and why it's fixing it. Something related to 64 bits ?
>>
>> And yes, it's never called ! I added this S/R in case one (like me,
>> for checking) wants or needs to use the matlab-topology-generator,
>> and then it will be called (by
>> utils/exch2/matlab-topology-generator/w2_e2setup.F)
>>
>> Could you try:
>>>     CHARACTER*1 edge(4), cc1(2)
>> and replace cc1 by cc1(1) in the 3 places where it is used
>> (line 82, 83 & 86 in w2_print_e2setup.F).
>>
>> Cheers,
>> Jean-Michel
>>
>> On Wed, Jun 03, 2009 at 11:58:33AM +0200, Martin Losch wrote:
>>> Sorry forgot to say (except in the email subject) that this happens  
>>> on
>>> my Intel AppleBookPro with Leopard.
>>>
>>> Further I noticed that this routine w2_print_e2setup is never  
>>> called, so
>>> that I have to suspect that it was only introduced to check whether  
>>> I am
>>> still running testreport on my laptop (o:
>>>
>>> Martin
>>>
>>> On Jun 3, 2009, at 11:18 AM, Martin Losch wrote:
>>>
>>>> Hi there, I am getting an error when running "testreport -t
>>>> global_ocean.cs32x15":
>>>>
>>>> g77 -Wunused -ffloat-store  -O0 -c w2_print_e2setup.f
>>>> w2_print_e2setup.f: In subroutine `w2_print_e2setup':
>>>> In file included from w2_print_e2setup.f:0:
>>>> w2_print_e2setup.f:1060: internal compiler error: in
>>>> instantiate_virtual_regs_lossage, at function.c:3746
>>>> Please submit a full bug report,
>>>> with preprocessed source if appropriate.
>>>> See <URL:http://gcc.gnu.org/bugs.html> for instructions.
>>>> make: *** [w2_print_e2setup.o] Error 1
>>>> make: Target `all' not remade because of errors.
>>>>
>>>> And this is the statement (appears twice) that causes this error (at
>>>> least the error goes away, when I comment it out):
>>>>          IF ( ii.NE.0 ) cc1 = edge(ii)
>>>> I can make it  compile by changing
>>>>     CHARACTER*1 edge(4), cc1
>>>> to
>>>>     CHARACTER*1 edge(4)
>>>>     CHARACTER*2 cc1
>>>> which is certainly not what the programmer intended, but maybe it  
>>>> give
>>>> you a hint, why this is happening and how it can be fixed.
>>>>
>>>> Martin
>>>> PS. I think g77 is no longer continued so I am not sure if it makes
>>>> sense to send a bug report
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> MITgcm-devel mailing list
>>>> MITgcm-devel at mitgcm.org
>>>> http://mitgcm.org/mailman/listinfo/mitgcm-devel
>>>
>>> _______________________________________________
>>> MITgcm-devel mailing list
>>> MITgcm-devel at mitgcm.org
>>> http://mitgcm.org/mailman/listinfo/mitgcm-devel
>> _______________________________________________
>> MITgcm-devel mailing list
>> MITgcm-devel at mitgcm.org
>> http://mitgcm.org/mailman/listinfo/mitgcm-devel
>
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel