CMake and Fortran compiler

[Mojca Miklavec]

On Sat, Dec 10, 2011 at 01:21, Chris Jones wrote:
>
> On 10 Dec 2011, at 12:02am, Mojca Miklavec wrote:
>
>> On Sat, Dec 10, 2011 at 00:45, Chris Jones wrote:
>>> On 9 Dec 2011, at 11:33pm, Mojca Miklavec wrote:
>>>
>>>> Hello,
>>>>
>>>> I'm trying to compile Root with CMake (the Root package is MacPorts is
>>>> broken, but even if it wasn't, there might be reasons why I would want
>>>> to use a standalone version).
>>>
>>> In what way is ROOT broken for you ?
>>
>> See my attachment in http://trac.macports.org/ticket/32477.
>
> and if you read further down you will find the problem.

... which is probably a different one than mine. (But nevertheless ...
when I wrote the email I didn't see the posts sent after I attached my
logs.)

> My bet is like another user you are using an external (and broken) gfortran compiler.

My bet is I'm not. Because "gfortran" doesn't exist on my machine and
also because the logs differ significantly from those by the other
user. Also, as I just figured out, the logs seems to suggest that root
doesn't find the fortran compiler at all.

:info:build ==> lib/libGenetic.so done
:info:build ==> lib/libRooFit.so done
:info:build /Developer/usr/bin/llvm-g++-4.2 -dynamiclib -single_module
-Wl,-dead_strip_dylibs -install_name
/opt/local/lib/root/libRooStats.so -O2 -m64 -mmacosx-version-min=10.7
-o lib/libRooStats.so roofit/roostats/src/AsymptoticCalculator.o
roofit/roostats/src/BayesianCalculator.o
roofit/roostats/src/BernsteinCorrection.o
roofit/roostats/src/ConfidenceBelt.o
roofit/roostats/src/FeldmanCousins.o
roofit/roostats/src/FrequentistCalculator.o
roofit/roostats/src/HLFactory.o roofit/roostats/src/Heaviside.o
roofit/roostats/src/HybridCalculator.o
roofit/roostats/src/HybridCalculatorOriginal.o
roofit/roostats/src/HybridPlot.o roofit/roostats/src/HybridResult.o
roofit/roostats/src/HypoTestCalculatorGeneric.o
roofit/roostats/src/HypoTestInverter.o
roofit/roostats/src/HypoTestInverterOriginal.o
roofit/roostats/src/HypoTestInverterPlot.o
roofit/roostats/src/HypoTestInverterResult.o
roofit/roostats/src/HypoTestPlot.o
roofit/roostats/src/HypoTestResult.o
roofit/roostats/src/LikelihoodInterval.o
roofit/roostats/src/LikelihoodIntervalPlot.o
roofit/roostats/src/MCMCCalculator.o
roofit/roostats/src/MCMCInterval.o
roofit/roostats/src/MCMCIntervalPlot.o
roofit/roostats/src/MarkovChain.o
roofit/roostats/src/MetropolisHastings.o
roofit/roostats/src/ModelConfig.o
roofit/roostats/src/NeymanConstruction.o
roofit/roostats/src/NumberCountingPdfFactory.o
roofit/roostats/src/NumberCountingUtils.o
roofit/roostats/src/PdfProposal.o
roofit/roostats/src/PointSetInterval.o
roofit/roostats/src/ProfileInspector.o
roofit/roostats/src/ProfileLikelihoodCalculator.o
roofit/roostats/src/ProfileLikelihoodTestStat.o
roofit/roostats/src/ProposalHelper.o
roofit/roostats/src/RatioOfProfileLikelihoodsTestStat.o
roofit/roostats/src/RooStatsUtils.o roofit/roostats/src/SPlot.o
roofit/roostats/src/SamplingDistPlot.o
roofit/roostats/src/SamplingDistribution.o
roofit/roostats/src/SequentialProposal.o
roofit/roostats/src/SimpleInterval.o
roofit/roostats/src/SimpleLikelihoodRatioTestStat.o
roofit/roostats/src/ToyMCSampler.o roofit/roostats/src/ToyMCStudy.o
roofit/roostats/src/UniformProposal.o
roofit/roostats/src/UpperLimitMCSModule.o
roofit/roostats/src/G__RooStats.o -ldl -Llib -lRooFit -lRooFitCore
-lTree -lRIO -lHist -lMatrix -lMathCore -lMinuit -lFoam -lGraf -lGpad
-Llib -lCore -lCint
:info:build Undefined symbols for architecture x86_64:
:info:build   "___GLXEW_ARB_multisample", referenced from:
:info:build       TGLFormat::InitAvailableSamples()      in TGLFormat.o
:info:build ld: symbol(s) not found for architecture x86_64
:info:build collect2: ld returned 1 exit status
:info:build make: *** [lib/libRGL.so] Error 1
:info:build make: *** Waiting for unfinished jobs....

It seems more like a failed llvm-g++-4.2 than failed fortran compiler,
but it is not excluded that this fails exactly because of the missing
compiler. (It is funny though that compiling root 5.30 worked out of
the box.)

> Try building with one of the gcc variants, to pick up a working one.

I will. Thank you for the hint.

But my point still stands, even if the version of Root suddenly starts
working. I would like to figure out what to do about missing Fortran
settings in MacPorts' CMake.

Mojca