Re: rootcint passing wrong options to cint?

Hi Philippe,

 > I can not reproduce the problem.

Neither can I locally. That's the stupid thing with the GRID, especially when there is only one site holding the complete dataset.

 > To debug this a bit further could you try (to find a way) to run  > rootcint with an addition debug

I'll try and let you know.

cu,
Alexander

Am 29.09.2010 16:45, schrieb Philippe Canal:
>
> Hi Alexander,
>
> I can not reproduce the problem.
>
> To debug this a bit further could you try (to find a way) to run
> rootcint with an addition debug
> flag (-v) placed just before the -p for example:
>
> /mnt/lustre_0/software/lcg_experimental_sw/sl5/atlas/prod/releases/rel_15-23/DetCommon/15.8.0/InstallArea/i686-slc5-gcc43-opt/bin/../../../../../sw/lcg/app/releases/ROOT/5.22.00j/i686-slc5-gcc43-opt/root/bin/rootcint
> -f src/SFrameCore_Dict.cxx -c -v -p
> -I/mnt/lustre_0/software/lcg_experimental_sw/sl5/atlas/prod/releases/rel_15-23/sw/lcg/app/releases/ROOT/5.22.00j/i686-slc5-gcc43-opt/root/include/
> -I/data/scratch/tmp/7834434.1.lcg_long/condorg_qJX22384/pilot3/Panda_Pilot_22523_1285763846/PandaJob_1121044084_1285763846/workDir/SFrame
> -I./ include/ISCycleBase.h include/ISCycleBaseConfig.h
> include/ISCycleBaseHist.h include/ISCycleBaseNTuple.h
> include/SConstants.h include/SCycleBase.h include/SCycleBaseBase.h
> include/SCycleBaseConfig.h include/SCycleBaseExec.h
> include/SCycleBaseHist.h include/SCycleBaseNTuple.h
> include/SCycleConfig.h include/SCycleController.h include/SCycleOutput.h
> include/SCycleStatistics.h include/SError.h include/SErrorHandler.h
> include/SFileMerger.h include/SGeneratorCut.h include/SInputData.h
> include/SLogWriter.h include/SLogger.h include/SMsgType.h
> include/SOutputFile.h include/SParLocator.h include/SPointer.h
> include/SProofManager.h include/SFrameCore_LinkDef.h
>
> Thanks,
> Philippe.
>
> On 9/29/10 8:30 AM, Alexander Mann wrote:
>>
>> Dear ROOT experts,
>>
>> I have the following problem when using rootcint (on the GRID, of
>> course). I am (well, actually the GRID is) calling rootcint like this:
>>
>> /mnt/lustre_0/software/lcg_experimental_sw/sl5/atlas/prod/releases/rel_15-23/DetCommon/15.8.0/InstallArea/i686-slc5-gcc43-opt/bin/../../../../../sw/lcg/app/releases/ROOT/5.22.00j/i686-slc5-gcc43-opt/root/bin/rootcint
>> -f src/SFrameCore_Dict.cxx -c -p
>> -I/mnt/lustre_0/software/lcg_experimental_sw/sl5/atlas/prod/releases/rel_15-23/sw/lcg/app/releases/ROOT/5.22.00j/i686-slc5-gcc43-opt/root/include/
>> -I/data/scratch/tmp/7834434.1.lcg_long/condorg_qJX22384/pilot3/Panda_Pilot_22523_1285763846/PandaJob_1121044084_1285763846/workDir/SFrame
>> -I./ include/ISCycleBase.h include/ISCycleBaseConfig.h
>> include/ISCycleBaseHist.h include/ISCycleBaseNTuple.h
>> include/SConstants.h include/SCycleBase.h include/SCycleBaseBase.h
>> include/SCycleBaseConfig.h include/SCycleBaseExec.h
>> include/SCycleBaseHist.h include/SCycleBaseNTuple.h
>> include/SCycleConfig.h include/SCycleController.h
>> include/SCycleOutput.h include/SCycleStatistics.h include/SError.h
>> include/SErrorHandler.h include/SFileMerger.h include/SGeneratorCut.h
>> include/SInputData.h include/SLogWriter.h include/SLogger.h
>> include/SMsgType.h include/SOutputFile.h include/SParLocator.h
>> include/SPointer.h include/SProofManager.h include/SFrameCore_LinkDef.h
>>
>> which gives an error message saying
>>
>> Error: Unknown option -opt/root/src
>> Error: rootcint: failed to open src/SFrameCore_Dict.h in
>> ReplaceBundleInDict()
>>
>> Looks like part of some file name is parsed as an option instead.
>> (Note that I am not passing any option or path to rootcint which
>> contains "root/src", so it must be taken from some configuration / env
>> variable.)
>>
>> I already found out that this message is not raised by rootcint
>> itself, but by cint. So somehow rootcint must pass a garbled option to
>> cint.
>>
>> You can find the full error log on
>>
>> http://voatlas20.cern.ch:25980/monitor/logs/9c0f3f29-28d0-49f9-8237-9eedffded1c2/tarball_PandaJob_1121044084_ANALY_QMUL/prun_stdout.txt
>>
>>
>> Locally the make process runs fine, as it also does on other sites,
>> see e.g.
>>
>> http://voatlas20.cern.ch:25980/monitor/logs/f26ae764-7dcb-4bf9-89bd-e68893c11612/tarball_PandaJob_1121044080_ANALY_INFN-T1/prun_stdout.txt
>>
>>
>> (which actually stems from the same job, so it is exactly the same
>> source code and makefiles.)
>>
>> cu,
>> Alexander
>>
Received on Wed Sep 29 2010 - 16:59:54 CEST