[Users] 'Ambiguous' g++ collisions on isnan (cmath), -std=gnu++0x

Fri Aug 3 12:01:16 CDT 2012

isnan has been around since about 1977. The fact that std::isnan
doesn't work with the C++ Intel compiler is either a compiler bug, or
an installation problem. Of course, many people won't realise this
until they try to build Cactus, so we have to provide a work-around.

The problem is that things probably depend on whether math.h or cmath
have been included, and in which order they have been included. C
requires isnan to be a macro, whereas C++ requires it to be a
function. Our current solution allows isnan to be inlined, which is
probably not so important for isnan, but may be important for signbit
or copysign. I wonder how a portable implementation of an (inlineable)
cctk_copysign would look like. It would mean that we stick the
work-arounds for compiler bugs and "misunderstanding" about the C++
standard into this function's implementation instead of option lists.

I think this is a good idea. cctk_math.h?

-erik

On Fri, Aug 3, 2012 at 12:08 PM, Ian Hinder <ian.hinder at aei.mpg.de> wrote:
>
> On 3 Aug 2012, at 13:45, Erik Schnetter wrote:
>
>> I don't know why the __builtin_isnan etc. are not defined for some
>> versions of the Intel compilers. I suspect an incompatibility with the
>> Intel compiler and the respective system libraries (which provide
>> math.h). I have updated the Simfactory option lists.
>
> The tests pass again (http://damiana2.aei.mpg.de/~ianhin/testreports/EinsteinToolkitTests/results.xml); thanks!
>
> Is there a more general way of solving this problem, rather than solving it in each machine's option list?  I'm worried that we have a lot of special-case options in our option lists now, and that new users will be essentially unable to use Cactus unless they have an expert to write an optionlist for them.  We have many exceptions and special flags in the optionlists now, each added to solve a specific problem.  I think these should be solved in a more general way.  For example, if isnan is not sufficiently portable that it can be used on any compiler, can we provide an autoconf test for it and a Cactus isnan function which calls the right thing?  This goes for all the other options which we have.  For example, for Datura, one currently has:
>
> CPPFLAGS =
> FPPFLAGS = -traditional
> CFLAGS   = -g -debug all -traceback -align -std=c99 -ansi_alias -U__STRICT_ANSI__ -rdynamic
> CXXFLAGS = -g -debug all -traceback -align -restrict -rdynamic -D__builtin_isnan=::isnan -D__builtin_signbit=::signbit
> F77FLAGS = -g -debug all -traceback -align -pad -w95 -cm
> F90FLAGS = -g -debug all -traceback -align -pad -w95 -cm
>
> I understand that adding these to the option list is the safest thing to do and provides a conservative quick fix, but at some point, shouldn't these things move into Cactus?
>
>
>>
>> -erik
>>
>> On Fri, Aug 3, 2012 at 3:44 AM, Ian Hinder <ian.hinder at aei.mpg.de> wrote:
>>>
>>> On 2 Aug 2012, at 22:40, Erik Schnetter wrote:
>>>
>>>> Try now.
>>>
>>> This causes a build failure on Datura:
>>>
>>>        http://git.barrywardell.net/EinsteinToolkitTestResults.git/blob/c32794e5e57db36da2fede2f1eec451daf108a9c:/build.log
>>>
>>>> /home/ianhin/Cactus/EinsteinToolkitTests/configs/EinsteinToolkitTests_bb930fc2b4a2900b3c5a780e79244bb0ddefb81d/lib/libthorn_Vectors.a(test.cc.o): In function `Vectors_Test':
>>>> /home/ianhin/Cactus/EinsteinToolkitTests/arrangements/LSUThorns/Vectors/src/test.cc:252: undefined reference to `__builtin_isnan'
>>>> /home/ianhin/Cactus/EinsteinToolkitTests/arrangements/LSUThorns/Vectors/src/test.cc:252: undefined reference to `__builtin_isnan'
>>>> /home/ianhin/Cactus/EinsteinToolkitTests/arrangements/LSUThorns/Vectors/src/test.cc:251: undefined reference to `__builtin_isnan'
>>>> /home/ianhin/Cactus/EinsteinToolkitTests/arrangements/LSUThorns/Vectors/src/test.cc:251: undefined reference to `__builtin_isnan'
>>>> /home/ianhin/Cactus/EinsteinToolkitTests/arrangements/LSUThorns/Vectors/src/test.cc:249: undefined reference to `__builtin_isnan'
>>>> /home/ianhin/Cactus/EinsteinToolkitTests/configs/EinsteinToolkitTests_bb930fc2b4a2900b3c5a780e79244bb0ddefb81d/lib/libthorn_Vectors.a(test.cc.o):/home/ianhin/Cactus/EinsteinToolkitTests/arrangements/LSUThorns/Vectors/src/test.cc:249: more undefined references to `__builtin_isnan' follow
>>>> make[1]: *** [/home/ianhin/Cactus/EinsteinToolkitTests/exe/cactus_EinsteinToolkitTests_bb930fc2b4a2900b3c5a780e79244bb0ddefb81d] Error 1
>>>> make: *** [EinsteinToolkitTests_bb930fc2b4a2900b3c5a780e79244bb0ddefb81d] Error 2
>>>> 0.10user 0.27system 12:11.56elapsed 0%CPU (0avgtext+0avgdata 14928maxresident)k
>>>> 0inputs+3032outputs (0major+1354minor)pagefaults 0swaps
>>>> Build failed
>>>
>>>>
>>>> -erik
>>>>
>>>> On 2012-08-02, at 15:36 , Erik Schnetter <schnetter at cct.lsu.edu> wrote:
>>>>
>>>>> The solution is likely to wrap the calls to isnan into a my_isnan function, similar to the way in which signbit is wrapped. I am testing a patch.
>>>>>
>>>>> -erik
>>>>>
>>>>> On 2012-08-02, at 15:14 , Scott Hawley <scott.hawley at belmont.edu> wrote:
>>>>>
>>>>>> I added '-D__builtin_isinf=::isinf -D__builtin_isnan=::isnan' as I saw in
>>>>>> a few simfactory .cfg files, but no luck.
>>>>>>
>>>>>> (Thanks to
>>>>>> http://lists.einsteintoolkit.org/pipermail/users/2011-June/001151.html)
>>>>>>
>>>>>> There is currently no nautilus.cfg file, but if someone would like access
>>>>>> to Nautilus in order to make one (which is what I seem to be doing,
>>>>>> slowly), I'd be happy to add you to my allocation! :-)
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> On 8/2/12 1:58 PM, "Scott Hawley" <scott.hawley at belmont.edu> wrote:
>>>>>>
>>>>>>> Hi.  The following is a well-documented collision that occurs with the
>>>>>>> cmath header.
>>>>>>>
>>>>>>> The fixes I've seen around the 'net all involve modifying the source and
>>>>>>> putting in #define statements. But the code in question isn't mine.
>>>>>>>
>>>>>>> Is there a "-f_____" compiler flag I set instead?
>>>>>>>
>>>>>>> Thanks,
>>>>>>> Scott
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> COMPILING
>>>>>>> /nics/c/home/hawleys/Cactus/arrangements/LSUThorns/Vectors/src/test.cc
>>>>>>> current_wd=`pwd` ; cd /nics/c/home/hawleys/Cactus/configs/texmex/scratch ;
>>>>>>> g++ -LMPICH_IGNORE_CXX_SEEK -g -DCARPET_DEBUG  -fopenmp  -O2 -m64
>>>>>>> -std=gnu++0x -g -O0 -O3 -fopenmp  -c -o $current_wd/test.cc.o
>>>>>>> $current_wd/test.cc
>>>>>>> -I"/nics/c/home/hawleys/Cactus/arrangements/LSUThorns/Vectors/src"
>>>>>>> -I"/nics/c/home/hawleys/Cactus/arrangements/LSUThorns/Vectors/src/include"
>>>>>>> -I"/nics/c/home/hawleys/Cactus/configs/texmex/config-data"
>>>>>>> -I"/nics/c/home/hawleys/Cactus/configs/texmex/bindings/include"
>>>>>>> -I"/nics/c/home/hawleys/Cactus/src/include"
>>>>>>> -I"/nics/c/home/hawleys/Cactus/arrangements"
>>>>>>> -I"/nics/c/home/hawleys/Cactus/configs/texmex/bindings/Configuration/Thorn
>>>>>>> s
>>>>>>> " -I"/nics/c/home/hawleys/Cactus/configs/texmex/bindings/include/Vectors"
>>>>>>> -I"/nics/c/home/hawleys/Cactus/arrangements/LSUThorns/Vectors/src"
>>>>>>> -I"/nics/c/home/hawleys/Cactus/arrangements/LSUThorns/Vectors/src/include"
>>>>>>> -I"/nics/c/home/hawleys/Cactus/configs/texmex/bindings/include/Vectors"
>>>>>>> -DCCODE
>>>>>>> /nics/c/home/hawleys/Cactus/configs/texmex/build/Vectors/test.cc: In
>>>>>>> function Œvoid Vectors_Test(cGH*)¹:
>>>>>>> /nics/c/home/hawleys/Cactus/configs/texmex/build/Vectors/test.cc:112:3:
>>>>>>> error: call of overloaded Œisnan(double&)¹ is ambiguous
>>>>>>> /nics/c/home/hawleys/Cactus/configs/texmex/build/Vectors/test.cc:112:3:
>>>>>>> note: candidates are:
>>>>>>> /usr/include/bits/mathcalls.h:235:1: note: int isnan(double)
>>>>>>> /nics/e/sw/local/nautilus/opt/gcc/4.6.3/bin/../lib/gcc/x86_64-unknown-linu
>>>>>>> x
>>>>>>> -gnu/4.6.3/../../../../include/c++/4.6.3/cmath:552:3: note: bool
>>>>>>> std::isnan(long double)
>>>>>>> /nics/e/sw/local/nautilus/opt/gcc/4.6.3/bin/../lib/gcc/x86_64-unknown-linu
>>>>>>> x
>>>>>>> -gnu/4.6.3/../../../../include/c++/4.6.3/cmath:548:3: note: bool
>>>>>>> std::isnan(double)
>>>>>>> /nics/e/sw/local/nautilus/opt/gcc/4.6.3/bin/../lib/gcc/x86_64-unknown-linu
>>>>>>> x
>>>>>>> -gnu/4.6.3/../../../../include/c++/4.6.3/cmath:544:3: note: bool
>>>>>>> std::isnan(float)
>>>>>>> /nics/c/home/hawleys/Cactus/configs/texmex/build/Vectors/test.cc:112:3:
>>>>>>> error: call of overloaded Œisnan(double&)¹ is ambiguous
>>>>>>> /nics/c/home/hawleys/Cactus/configs/texmex/build/Vectors/test.cc:112:3:
>>>>>>> note: candidates are:
>>>>>>> /usr/include/bits/mathcalls.h:235:1: note: int isnan(double)
>>>>>>> (  ...more of the same...   )
>>>>>>> make[3]: *** [test.cc.o] Error 1
>>>>>>> make[3]: Leaving directory
>>>>>>> `/nics/c/home/hawleys/Cactus/configs/texmex/build/Vectors'
>>>>>>> make[2]: *** [make.checked] Error 2
>>>>>>> make[2]: Leaving directory
>>>>>>> `/nics/c/home/hawleys/Cactus/configs/texmex/build/Vectors'
>>>>>>> make[1]: ***
>>>>>>> [/nics/c/home/hawleys/Cactus/configs/texmex/lib/libthorn_Vectors.a] Error
>>>>>>> 2
>>>>>>> make[1]: Leaving directory `/nics/c/home/hawleys/Cactus'
>>>>>>> make: *** [texmex] Error 2
>>>>>>>
>>>>>>> hawleys/Cactus> g++ --version
>>>>>>> g++ (GCC) 4.6.3
>>>>>>> Copyright (C) 2011 Free Software Foundation, Inc.
>>>>>>> This is free software; see the source for copying conditions.  There is NO
>>>>>>> warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR
>>>>>>> PURPOSE.
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> Users mailing list
>>>>>>> Users at einsteintoolkit.org
>>>>>>> http://lists.einsteintoolkit.org/mailman/listinfo/users
>>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> Users mailing list
>>>>>> Users at einsteintoolkit.org
>>>>>> http://lists.einsteintoolkit.org/mailman/listinfo/users
>>>>>
>>>>> --
>>>>> Erik Schnetter <eschnetter at perimeterinstitute.ca>
>>>>> http://www.perimeterinstitute.ca/personal/eschnetter/
>>>>> AIM: eschnett247, Skype: eschnett, Google Talk: schnetter at gmail.com
>>>>>
>>>>> _______________________________________________
>>>>> Users mailing list
>>>>> Users at einsteintoolkit.org
>>>>> http://lists.einsteintoolkit.org/mailman/listinfo/users
>>>>
>>>> --
>>>> Erik Schnetter <eschnetter at perimeterinstitute.ca>
>>>> http://www.perimeterinstitute.ca/personal/eschnetter/
>>>> AIM: eschnett247, Skype: eschnett, Google Talk: schnetter at gmail.com
>>>>
>>>> _______________________________________________
>>>> Users mailing list
>>>> Users at einsteintoolkit.org
>>>> http://lists.einsteintoolkit.org/mailman/listinfo/users
>>>
>>> --
>>> Ian Hinder
>>> http://numrel.aei.mpg.de/people/hinder
>>>
>>
>>
>>
>> --
>> Erik Schnetter <schnetter at cct.lsu.edu>
>> http://www.perimeterinstitute.ca/personal/eschnetter/
>
> --
> Ian Hinder
> http://numrel.aei.mpg.de/people/hinder
>

-- 
Erik Schnetter <schnetter at cct.lsu.edu>
http://www.perimeterinstitute.ca/personal/eschnetter/