[Rivet] Rivet on afs

[Andy Buckley]

Thanks for the info, Kathrin! I'm glad it's working now... phew!

I've updated the README file in our "experimental" directory, and we
should put this info in the Rivet FAQ page, too.

General "LCG environment" setup scripts would still be useful, I think,
for the reasons I gave above, but that's not Kathrin's problem :-)

Cheers,
Andy


On 17/11/14 12:05, Kathrin Becker wrote:
> Dear Dimitri, dear Andy,
> 
> thank you for your fast response!
> As suspected by Andy, I used setupATLAS to change the gcc and python
> version.
> If I use the rivetenv-genser.sh script, that Dimitri suggests,
> everything runs fine.
> Also if I use:
> 
> setupATLAS
> localSetupGcc gcc472_x86_64_slc6
> localSetupPython 2.7.3-x86_64-slc6-gcc47
> source /afs/cern.ch/sw/lcg/external/MCGenerators_lcgcmt67c/rivet/2.2.0/x86_64-slc6-gcc47-opt/rivetenv.sh
> 
> it runs without errors.
> Cheers,
> 
> Kathrin
> 
> PS: Next time I'll write to the list directly. Sorry about the spam!
> 
> 
> 
> 
> On 11/17/2014 01:01 PM, Andy Buckley wrote:
>> On 17/11/14 10:50, Dmitri Konstantinov wrote:
>>> Hi Andy,
>>>
>>> For GENSER we use LCG setup for gcc environment, i.e.
>>>
>>>              source
>>> /afs/cern.ch/sw/lcg/external/gcc/4.7.2/x86_64-slc6-gcc47-opt/setup.sh
>> But this doesn't set up Python, so another sourcing would be needed for
>> that. And maybe there are other elements of the LCG platform definition
>> that are also needed? It's not obvious to "average users" (or to me)
>> what the right combinations are, or whether the sourcing scripts need to
>> be called in a particular order. It also requires some exploration of
>> the LCG AFS external area to find the hopefully correct setup scripts.
>> It would be trivial, but very helpful, if there were "master" setup
>> scripts for all supported LCG environments in e.g.
>> /afs/cern.ch/sw/lcg/envsetup/
>>
>>> I think it is question for Emil - what is  difference between ATLAS and
>>> LCG gcc environment. or what is proper way to set ATLAS gcc environment?
>> I was under the impression that "localSetupPython
>> 2.7.3-x86_64-slc6-gcc47" had also set up gcc47, but on re-testing I see
>> that it doesn't. Maybe the issue is related to that, since gcc 4.4 and
>> 4.7 are binary compatible, but maybe libquadmath is a special case,
>> given what it does. @Kathrin, does it work if you add a localSetupGcc
>> line, like this?
>>
>> setupATLAS
>> localSetupGcc gcc472_x86_64_slc6
>> localSetupPython 2.7.3-x86_64-slc6-gcc47
>> source
>> /afs/cern.ch/sw/lcg/external/MCGenerators_lcgcmt67c/rivet/2.2.0/x86_64-slc6-gcc47-opt/rivetenv.sh
>>
>> Thanks,
>> Andy
>>
>>
>>> On 11/17/14 3:40 PM, Andy Buckley wrote:
>>>> Hi Dmitri,
>>>> In general it would be very useful to have general LCG setup scripts so
>>>> I can do, for example, "source /path/to/lcg_gcc47_python27.sh" (one for
>>>> each supported LCG platform) and get the equivalent of the environment
>>>> that you use to do your builds. I have mentioned this once or twice
>>>> before :-P
>>> I have mentioned in my previous mail that we have such kind of
>>> environment script for every generator/package not for whole LCG
>>> release. ;)
>>>
>>> If you would like set necessary environment for RIVET - just run:
>>>
>>> source
>>> /afs/cern.ch/sw/lcg/external/MCGenerators_lcgcmt67c/rivet/2.2.0/x86_64-slc6-gcc47-opt/rivetenv-genser.sh
>>>
>>>
>>> Cheers,
>>>     Dima
>>>
>>>
>>>> Thanks,
>>>> Andy
>>>>
>>>>
>>>> On 17/11/14 10:21, Dmitri Konstantinov wrote:
>>>>> Hello Andy, Hello Katrhin,
>>>>>
>>>>> I would propose to use our GENSER environment scripts provided for each
>>>>> package.
>>>>>
>>>>> source
>>>>> /afs/cern.ch/sw/lcg/external/MCGenerators_lcgcmt67c/rivet/2.2.0/x86_64-slc6-gcc47-opt/rivetenv-genser.sh
>>>>>
>>>>>
>>>>>
>>>>> I think Kathrin doesn't have properly defined gcc environment for given
>>>>> gcc versions - gcc47, while default(system) lxplus compiler is gcc 4.4
>>>>>
>>>>> Cheers,
>>>>>       Dima
>>>>>
>>>>>
>>>>>
>>>>> On 11/17/14 3:13 PM, Andy Buckley wrote:
>>>>>> Hi Kathrin,
>>>>>>
>>>>>> I have copied this to the Rivet developer mailing list -- please send
>>>>>> queries there rather than to me.
>>>>>>
>>>>>> I'm not sure about this issue -- it looks to me like the Genser
>>>>>> build of
>>>>>> Rivet/YODA was made using a different build of GCC than the one
>>>>>> available via setupATLAS. We do not use libquadmath at all, so it is
>>>>>> something that has been linked in by default.
>>>>>>
>>>>>> I suspect that finding a copy of libquadmath on AFS and putting its
>>>>>> containing directory into your LD_LIBRARY_PATH variable would solve the
>>>>>> problem, but there should be a standard environment setup for using LCG
>>>>>> builds without having to do manual tweaks like that. I've copied in the
>>>>>> LCG application area manager and Generator Services team, who I hope
>>>>>> can
>>>>>> provide an answer. I would also like to know, so I can provide easy
>>>>>> instructions to use Genser's Rivet build, for non-ATLAS users.
>>>>>>
>>>>>> Best wishes,
>>>>>> Andy
>>>>>>
>>>>>>
>>>>>> On 17/11/14 10:52, Kathrin Becker wrote:
>>>>>>> Dear Andy,
>>>>>>>
>>>>>>> I'm sorry to bother you again. I tried today to switch away from the
>>>>>>> experimental version of rivet to the other one in afs.
>>>>>>> I followed your instructions in the email.
>>>>>>> But if I now do rivet-cmphistos I get the following error.
>>>>>>> [becker at pplxint9 valplots_WW]$ rivet-cmphistos
>>>>>>> mc12_8TeV.181416.PowhegPythia8_AU2CT10_HJ_MINLO_ggH125_WW2lep_EF_15_5.evgen.EVNT.e2177_tid01320008_00.yoda
>>>>>>>
>>>>>>>
>>>>>>> mc12_8TeV.161224.PowhegPythia8_AU2CT10_ggH125_WW2lep.evgen.EVNT.e2363_tid01352492_00.yoda
>>>>>>>
>>>>>>>
>>>>>>> 181999.HJ_MINLO_CT10nnlo_8TeV.yoda
>>>>>>> 181997.H_NNLO_22_11_CT10nnlo_8TeV.yoda less standaloneSetupRivet.sh
>>>>>>> Traceback (most recent call last):
>>>>>>>     File
>>>>>>> "/afs/cern.ch/sw/lcg/external/MCGenerators_lcgcmt67c/rivet/2.2.0/x86_64-slc6-gcc47-opt/bin/rivet-cmphistos",
>>>>>>>
>>>>>>>
>>>>>>> line 13, in <module>
>>>>>>>       import rivet, yoda, sys, os
>>>>>>>     File
>>>>>>> "/afs/cern.ch/sw/lcg/external/MCGenerators_lcgcmt67c/rivet/2.2.0/x86_64-slc6-gcc47-opt/lib/python2.7/site-packages/rivet/__init__.py",
>>>>>>>
>>>>>>>
>>>>>>> line 12, in <module>
>>>>>>>       from rivet.core import *
>>>>>>> ImportError: libquadmath.so.0: cannot open shared object file: No such
>>>>>>> file or directory
>>>>>>>
>>>>>>> Do you have an idea what I'm doing wrong?
>>>>>>> Thank you for your help!
>>>>>>> Cheers,
>>>>>>> Kathrin
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> On 11/12/2014 06:55 PM, Andy Buckley wrote:
>>>>>>>> On 12/11/14 16:42, Kathrin Becker wrote:
>>>>>>>>> Dear Andy,
>>>>>>>>>
>>>>>>>>> maybe you have already gotten tons of messages about this... But
>>>>>>>>> just in
>>>>>>>>> case you haven't:
>>>>>>>> Actually I didn't get lots of emails about this... you're the first.
>>>>>>>> Yes, I was surprised too!
>>>>>>>>
>>>>>>>>> Is there a reason why the rivet on afs has been move from
>>>>>>>>> /afs/cern.ch/sw/lcg/experimental/rivet/ to
>>>>>>>>> /afs/cern.ch/sw/lcg/experimental/rivet/bak/ ?
>>>>>>>>> Thus, all setup scripts don't work anymore.
>>>>>>>>> Should we use a different rivet, that is located somewhere else?
>>>>>>>> Yes, that's right! That area was never meant for production use --
>>>>>>>> the
>>>>>>>> hint is in the "experimental" bit of the path! The version in
>>>>>>>> there was
>>>>>>>> a pre-release version from about 2 years ago; and was subject to
>>>>>>>> change
>>>>>>>> at any time (I'll use it again if I ever need to make a slightly
>>>>>>>> public
>>>>>>>> pre-release copy available for testing.)
>>>>>>>>
>>>>>>>> I just added a README file to that directory explaining how to set
>>>>>>>> up a
>>>>>>>> reliable existing AFS Rivet build, but here's the short version:
>>>>>>>>
>>>>>>>> setupATLAS
>>>>>>>> localSetupPython 2.7.3-x86_64-slc6-gcc47
>>>>>>>> source
>>>>>>>> /afs/cern.ch/sw/lcg/external/MCGenerators_lcgcmt67c/rivet/2.2.0/x86_64-slc6-gcc47-opt/rivetenv.sh
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>> PS: I only use rivet outside of Athena to do make-plots, as this
>>>>>>>>> sometimes doesn't work when athena is setup.
>>>>>>>> Hmm, that's odd. How does it go wrong, and with what release(s)?
>>>>>>>>
>>>>>>>> Andy
>>>>>>>>
>>
> 


-- 
Dr Andy Buckley, Royal Society University Research Fellow
Particle Physics Expt Group, University of Glasgow / PH Dept, CERN