[Rivet] Fwd: Rivet on MAC

[Andy Buckley]

James reminded me last night that some time ago I tried to get Rivet to 
build with clang(++) on Ubuntu. It took some initial fiddling to get 
clang to recognise the position of the std libraries, and when it did 
find them my recollection is that it produced a lot of warnings: it 
looks like a *very* strict compiler (hmm, and they're using this to 
rewrite ROOT's CINT parser? Could be interesting...)

If clang has become the default compiler on Macs, I presume that they've 
sorted out the std library issues, but the strictness could still be an 
issue for Rivet and the libraries it depends on. Maybe there is a flag 
to tell it to be less obsessive, at least for now.

Andy


On 30/01/12 18:33, James Monk wrote:
> I haven't read all of that thoroughly, but I notice he mentions clang.  My understanding is that clang is the default compiler under 10.7.  In principle this is fine and should work with a GNU tool chain and be basically swappable for gcc, but I've never used 10.7 myself (other than the brief period when my new laptop came installed with it), so I don't know if that's actually the case (in fact this was one of the reasons I haven't "upgraded" to 10.7).
>
> So it isn't that Rivet "tries to compile itself... with clang," it doesn't have much of a choice about that.  The fact that autotools has obviously detected clang and tried to run it is actually encouraging.  I'd guess things might go a bit wrong if it  (or the user, via symlinks) tries to mix together compilers in the build, but I don't really know - somewhere back in London I have a 10.7 boot disk from which I could try all of this out, but right now I'm a bit busy with MC12 ;)
>
> cheers,
>
> James
>
>
> On 30 Jan 2012, at 15:57, David Grellscheid wrote:
>
>> Just found this... any input?
>>
>>   David
>>
>>
>> -------- Original Message --------
>> Subject: Rivet on MAC
>> Date: Sun, 15 Jan 2012 22:40:08 +0100
>> From: Juergen Reuter<juergen.reuter at desy.de>
>> To: David Grellscheid<david.grellscheid at durham.ac.uk>, andy.buckley at ed.ac.uk
>> CC: Daniel Wiesler<daniel.wiesler at desy.de>,        Jürgen Reuter<juergen.reuter at desy.de>
>>
>> Hi David,
>> I think you already have some tickets concerning Rivet under MAC OS X,
>> but nevertheless I summarize
>> my problems here:
>> *****
>> 1) even when configured with --without-yaml_cpp, as soon as the yaml_cpp
>> lib and headers are installed,
>> I get this error in AnalysisInfo. Uninstalling yaml_cpp (v0.2.6) solves
>> the problem.
>> CXX libRivetCore_la-AnalysisInfo.lo
>> AnalysisInfo.cc: In static member function ‘static Rivet::AnalysisInfo*
>> Rivet::AnalysisInfo::make(const string&)’:
>> AnalysisInfo.cc:67:28: error: ‘const class YAML::Node’ has no member
>> named ‘GetType’
>> AnalysisInfo.cc:67:41: error: ‘CT_SCALAR’ is not a member of ‘YAML’
>> AnalysisInfo.cc:68:28: error: ‘const class YAML::Node’ has no member
>> named ‘GetType’
>> AnalysisInfo.cc:68:41: error: ‘CT_SCALAR’ is not a member of ‘YAML’
>> AnalysisInfo.cc:77:25: error: ‘const class YAML::Node’ has no member
>> named ‘GetType’
>> AnalysisInfo.cc:77:38: error: ‘CT_SCALAR’ is not a member of ‘YAML’
>> AnalysisInfo.cc:78:25: error: ‘const class YAML::Node’ has no member
>> named ‘GetType’
>> AnalysisInfo.cc:78:38: error: ‘CT_SCALAR’ is not a member of ‘YAML’
>> AnalysisInfo.cc:94:21: error: ‘const class YAML::Node’ has no member
>> named ‘GetType’
>> AnalysisInfo.cc:94:34: error: ‘CT_SCALAR’ is not a member of ‘YAML’
>> AnalysisInfo.cc:99:28: error: ‘const class YAML::Node’ has no member
>> named ‘GetType’
>> AnalysisInfo.cc:99:41: error: ‘CT_SEQUENCE’ is not a member of ‘YAML’
>> ../../include/Rivet/Math/MathHeader.hh: At global scope:
>> ../../include/Rivet/Math/MathHeader.hh:42:16: warning: ‘Rivet::MAXINT’
>> defined but not used [-Wunused-variable]
>> *****
>> 2) When compiling the OPAL2004 Analysis I get the following error:
>> CXX OPAL_2004_S6132243.lo
>> OPAL_2004_S6132243.cc: In member function ‘virtual void
>> Rivet::OPAL_2004_S6132243::analyze(const Rivet::Event&)’:
>> OPAL_2004_S6132243.cc:167:25: error: ‘isnan’ was not declared in this scope
>> OPAL_2004_S6132243.cc:167:25: note: suggested alternatives:
>> /usr/local/lib/gcc/x86_64-apple-darwin11.2.0/4.6.2/../../../../include/c++/4.6.2/cmath:768:5:
>> note: ‘std::isnan’
>> /opt/local/include/boost/math/special_functions/fpclassify.hpp:515:25:
>> note: ‘boost::math::isnan’
>> ../../include/Rivet/Math/MathHeader.hh: At global scope:
>> ../../include/Rivet/Math/MathHeader.hh:42:16: warning: ‘Rivet::MAXINT’
>> defined but not used [-Wunused-variable]
>>
>> There is still a TODO in your code ;) but commenting out the isnan catch
>> solves the problem.
>> This is funny because the file cmath which contains the isnan is
>> identical to the one on a Linux machine (same gcc) where
>> I do not have a problem.
>> *****
>> 3) Next thing is that Rivet tries to compile itself in the (almost)
>> final step with clang(++) under MAC which utterly
>> (cf. extra file attached). This seems to be hard-coded. If I set a sym
>> link forcing clang to be gcc and clang++ to be
>> g++ it works.
>> *****
>> 4) There is also a problem in doc:
>> LD_LIBRARY_PATH=../src/.libs:/usr/local/lib:/usr/local/lib: \
>> DYLD_LIBRARY_PATH=../src/.libs:/usr/local/lib:/usr/local/lib: \
>> RIVET_INFO_PATH=../data/anainfo \
>> /opt/local/bin/python mk-analysis-latex
>> Traceback (most recent call last):
>> File "mk-analysis-latex", line 43, in<module>
>> import rivet
>> File
>> "/usr/local/packages/Rivet-1.7.0/pyext/build/lib.macosx-10.7-x86_64-2.7/rivet/__init__.py",
>> line 16, in<module>
>> from rivetwrap import *
>> File
>> "/usr/local/packages/Rivet-1.7.0/pyext/build/lib.macosx-10.7-x86_64-2.7/rivet/rivetwrap.py",
>> line 26, in<module>
>> _rivetwrap = swig_import_helper()
>> File
>> "/usr/local/packages/Rivet-1.7.0/pyext/build/lib.macosx-10.7-x86_64-2.7/rivet/rivetwrap.py",
>> line 22, in swig_import_helper
>> _mod = imp.load_module('_rivetwrap', fp, pathname, description)
>> ImportError:
>> dlopen(/usr/local/packages/Rivet-1.7.0/pyext/build/lib.macosx-10.7-x86_64-2.7/rivet/_rivetwrap.so,
>> 10): Library not loaded: /usr/local/lib/libyaml-cpp.0.dylib
>> Referenced from: /usr/local/packages/Rivet-1.7.0/src/.libs/libRivet.10.dylib
>> Reason: image not found
>>
>> Due to some reason, Rivet/dlopen believes to depend on some libraries in
>> /usr/local which do not exist!? This only happens for
>> libyaml_cpp. Problem here is that the package to be installed on MAC OS
>> X Lion only contains the static build, but Rivet demands
>> for the dynamic build. I solved the problem by manually copying
>> src/Tools/yaml-cpp/.libs/libyaml-cpp.0.dylib to $(prefix)/lib.
>>
>> Then Rivet compiled (!).
>> But running it immediatly gave a Python runtime error, I'm using Python
>> 2.7. :(((
>> I attached the crash report.
>> So I guess, it runs on Scientific Linux, but on my laptop is not usable.
>> Cheers,
>> JRR
>>
>> --
>> -------------------------------------
>> Juergen Reuter	
>> *****
>> DESY Theory Group
>> Notkestrasse 85
>> D-22603 Hamburg
>> Phone: +49 (0)40 8998-3895
>> Fax  : +49 (0)40 8998-2777
>>
>> *****
>> Skype: jr_reuter
>> -------------------------------------
>>
>>
>> <rivet_clang.txt><rivet_crash.txt><config.log>_______________________________________________
>> Rivet mailing list
>> Rivet at projects.hepforge.org
>> http://www.hepforge.org/lists/listinfo/rivet
>
> _______________________________________________
> Rivet mailing list
> Rivet at projects.hepforge.org
> http://www.hepforge.org/lists/listinfo/rivet
>


-- 
Dr Andy Buckley
SUPA Advanced Research Fellow
Particle Physics Experiment Group, University of Edinburgh

The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.