[Users] Summatiobyparts compilation failure on ARA cluster (latest ET release)

[Peter Diener]

Hi Andrea,

Roland pointed out in a separate e-mail that in your optionlist, you set 
FPP = fpp. I didn't notice that earlier. It is very likely that setting
FPP = cpp instead will fix your problem. This is used in all option 
files in simfactory.

Cheers,

   Peter

On Friday 2018-11-16 15:20, Peter Diener wrote:

>Date: Fri, 16 Nov 2018 15:20:56
>From: Peter Diener <diener at cct.lsu.edu>
>To: Roland Haas <rhaas at illinois.edu>
>Cc: Erik Schnetter <schnetter at protonmail.com>,
>    Andrea Endrizzi <aendrizzi7 at gmail.com>,
>    Einstein Toolkit Users <users at einsteintoolkit.org>
>Subject: Re: [Users] Summatiobyparts compilation failure on ARA cluster (latest
>     ET release)
>
> hi Roland,
>
> FPPFLAGS=-traditional
>
> was included in the optionlist that Andrea attached to the e-mail.
>
> Cheers,
>
>  Peter
>
> On Friday 2018-11-16 15:17, Roland Haas wrote:
>
>> Date: Fri, 16 Nov 2018 15:17:08
>> From: Roland Haas <rhaas at illinois.edu>
>> To: Peter Diener <diener at cct.lsu.edu>
>> Cc: Erik Schnetter <schnetter at protonmail.com>,
>>    Andrea Endrizzi <aendrizzi7 at gmail.com>,
>>    Einstein Toolkit Users <users at einsteintoolkit.org>
>> Subject: Re: [Users] Summatiobyparts compilation failure on ARA cluster 
>> (latest
>>     ET release)
>> 
>> Hello all,
>> 
>> this is just a blind guess but I have seen various strange Fortan
>> compiler errors if one forgets to include "-traditional" in the FPPFLAGS
>> options in the option list.
>> 
>> If you do not set FPPFLAGS at all then you should be good since the
>> default includes it, however if you eg have
>> 
>> FPPFLAGS=-DFOO
>> 
>> then you have to change it to be
>> 
>> FPPFLAGS=-DFOO -traditional
>> 
>> Yours,
>> Roland
>> 
>>> Hi Andrea,
>>> 
>>> The error message points to line 11 in the source file, which is the
>>> line that contains
>>> 
>>> DECLARE_CCTK_FUNCTIONS
>>> 
>>> This is expanded by a C-preprocessor before being handed of to the
>>> compiler.
>>> 
>>> The preprocessed file can be found in the directory:
>>> 
>>> /home/wa65fop/CS_Wu/THC_ZM1/Cactus/configs/<your config 
>>> name>/build/SummationByParts
>>> 
>>> and is named Derivatives_2_1.f90.
>>> 
>>> It seems like the C-preprocessor is not expanding
>>> "DECLARE_CCTK_FUNCTIONS" correctly and that the preprocessed file
>>> contains a line like:
>>>
>>>    end interface                                 &   interface
>>> 
>>> which is clearly not valid fortran. Can you check your preprocessed file
>>> and verify that this is indeed present?
>>> 
>>> I don't see anything obviously wrong with your preprocessor flags, but
>>> why is there a
>>> -I/opt/intel/compilers_and_libraries_2018.0.128/linux/mpi/intel64/include
>>> in your CPPFLAGS in you options file?
>>> 
>>> Cheers,
>>>
>>>    Peter
>>> 
>>> On Friday 2018-11-16 11:08, Erik Schnetter wrote:
>>> 
>>>> Date: Fri, 16 Nov 2018 11:08:48
>>>> From: Erik Schnetter <schnetter at protonmail.com>
>>>> To: Andrea Endrizzi <aendrizzi7 at gmail.com>
>>>> Cc: Einstein Toolkit Users <users at einsteintoolkit.org>
>>>> Subject: Re: [Users] Summatiobyparts compilation failure on ARA cluster 
>>>> (latest
>>>>     ET release)
>>>> 
>>>> Andrea
>>>> 
>>>> The error message quotes a line that is not and should not be present
>>>> in the source code. I wonder whether maybe your source file has
>>>> somehow become corrupted?
>>>> 
>>>> Can you attach the file
>>>> /home/wa65fop/CS_Wu/THC_ZM1/Cactus/arrangements/CactusNumerical/SummationByParts/src/Derivatives_2_1.F90
>>>> from your source tree, as well as the preprecessed file
>>>> /home/wa65fop/CS_Wu/THC_ZM1/Cactus/configs/whiskyTHC_ZM1/build/SummationByParts/Derivatives_2_1.f90?
>>>> 
>>>> -erik
>>>> 
>>>>> On Nov 16, 2018, at 1:51 , Andrea Endrizzi <aendrizzi7 at gmail.com> wrote:
>>>>> 
>>>>> Good morning,
>>>>>   I was trying to compile ET on ARA cluster, specifically with the
>>>>> latest intel compilers (2018, Update 3). All the C/C++ code compiled
>>>>> nicely, but when reaching the SummationByParts thorn, the file
>>>>> Derivatives_2_1.F90 failed (attached is a .txt of the error). I never
>>>>> saw this kind of error before when compiling previous versions of ET,
>>>>> but it seems to complain about syntax in a file that looked ok when I
>>>>> opened the source. Could it be related to the compiler/the compiler's
>>>>> flags I'm using? If that may be the case, I also attach the .cfg file
>>>>> I'm using (architecture of the nodes is Skylake). I am currently
>>>>> trying to build with a previous version, to check whether it is
>>>>> version-related or just a problem on my side.
>>>>> 
>>>>> Thanks already for your attention!
>>>>> 
>>>>> Best regards,
>>>>> 
>>>>> Andrea Endrizzi
>>>>> <error_report_summationbyparts.txt><ara-skl-icc_v0.1.cfg>
>>>> 
>>>> --
>>>> Erik Schnetter <schnetter at protonmail.com>
>>>> http://www.perimeterinstitute.ca/personal/eschnetter/
>>>> 
>>>> 
>>>> 
>>>> 
>>> 
>>> _______________________________________________
>>> Users mailing list
>>> Users at einsteintoolkit.org
>>> http://lists.einsteintoolkit.org/mailman/listinfo/users
>> 
>> 
>> 
>> -- 
>> My email is as private as my paper mail. I therefore support encrypting
>> and signing email messages. Get my PGP key from http://keys.gnupg.net.
>> 
>
>