conflict between libraries for MV and TPC with pax_v6.6.8

[mlotfiben]

When initialising both the TPC and MV one after the other I have encountered a problem that seems to do with the libraries that are responsible for reading root files into pandas data frames (I am not an expert on this, I may well be wrong here). I am attaching two files one which shows the error and one python file that can help in reproducing the error.
error_message.txt
`
import os
import numpy as np
import pandas as pd
from tqdm import tqdm
from scipy import stats, optimize, interpolate
from scipy.optimize import curve_fit

from scipy.misc import factorial
from glob import glob
import matplotlib
import matplotlib.pyplot as plt
import pickle
import hax
from hax import cuts
from pax import configuration, units, datastructure
pax_version = "pax_v6.8.0"
hax.init(experiment='XENON1T',
minitree_paths =['/project2/lgrandi/xenon1t/minitrees/%s/'%pax_version],
make_minitrees = False,
detector='tpc')

data = hax.minitrees.load([11345],["Basics","Fundamentals", "TotalProperties", "LargestPeakProperties","Corrections"])
print(len(data.index))

hax.init(experiment='XENON1T',
#minitree_paths =['/project2/lgrandi/xenon1t/minitrees/%s'%pax_version, '/project/lgrandi/xenon1t/minitrees/%s'%self.pax_version],
main_data_paths=['/project2/lgrandi/xenon1t/processed/%s/'%pax_version],
detector='muon_veto')

data_mv = hax.minitrees.load(["170718_0445"],treemakers=["MuonVetoTreemaker"])

data_mv.keys()
`

[mlotfiben]

An optimum, but not final solution was found with the help of Fei. Actually before I initialise the MV through hax, I need to delete the rootpy directory in .cache and also all libraries related to pax where I run my script. By the way this problem was not there when I used pax_V6.6.5.

[sanderbreur]

Dear Lotfi, Glad to hear the problem is solved. What you describe is not a problem of a specific pax version. This same problem has been encountered many times when for some reason the root class file is incorrectly written. It is just some of the root magic, and not specific for any pax version. Cheers, S

…

On Wed, 18 Oct 2017, 23:41 Mohamed Lotfi Benabderrahmane, < ***@***.***> wrote: An optimum, but not final solution was found with the help of Fei. Actually before I initialise the MV through hax, I need to delete the rootpy directory in .cache and also all libraries related to pax where I run my script. By the way this problem was not there when I used pax_V6.6.5. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#629 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AGNstTFzbhTYp-mBcKW_6uHAxbMgv_GRks5stnB9gaJpZM4P9-vT> .

[mlotfiben]

Hi Sander,
I am not sure if I consider this is a final solution because the "some reason" that you mentioned needs to be identified and if possible fixed.
Best,
Lotfi

[sanderbreur]

Dear Lotfi, Of course a fix would be great, the problem is that the amount of manpower/expertise neede is unclear. As we identified this problem last year already, the solution is stated clearly in the analysis guide. Any analyst who encounters it follows these steps to clear the problem. It is normally a one time fix. For the MV analysis this problem is now solved right? Or does it keep reoccurring every time you run? Cheers, S

…

On Thu, 19 Oct 2017, 10:27 Mohamed Lotfi Benabderrahmane, < ***@***.***> wrote: Hi Sander, I am not sure if I consider this is a final solution because the "some reason" that you mentioned needs to be identified and if possible fixed. Best, Lotfi — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#629 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AGNstchAl7KFqE_E6dUX0Z1c9koSJeQJks5stwgAgaJpZM4P9-vT> .

[JelleAalbers]

I think it crashes because the muon veto and TPC event classes have a different number of channels, causing a mismatch in the compiled classes. If I'm right you'll get this issue if you use pax for any two different detectors.

Still, this is odd since, from your error message, ROOT apparently does start recompiling something when you load MV minitrees (as it should: pax will ask ROOT to recompile stuff whenever you load a file with a different C++ class, even different in a small detail). Some magic is apparently lingering...

I agree cleaning cache files when switching from TPC->MV isn't ideal, but I don't see another way at the moment.