Fix crash in TupleToolHerschel when multiple versions are present
TupleToolHerschel causes a crash without any error message when a file from a run with one calibration is followed by a file from another run with another calibration in the same job. @crsanche ran into the problem with files 00069074_00008083_1.ew.dst and 00069074_00010274_1.ew.dst from 2015.
The reason is that the name of the variable is changed depending on calibration. On the first J/psi->mumu event of these two files we have
log_hrc_fom_v5 = 5.62789
log_hrc_fom_B_v5 = 5.61727
log_hrc_fom_F_v5 = 1.07714
and
log_hrc_fom_v4 = -15.9944
log_hrc_fom_B_v4 = -17.4217
log_hrc_fom_F_v4 = -16.2688
This is a bug. A TupleTool should not crash. I suggest to add a variable that gives the version and remove it from the name of the log chi2.
Merge request reports
Activity
added bug fix label
requested review from @johndan
assigned to @masmith
- [2022-02-04 00:41] Validation started with lhcb-run2-gaudi-head#273
- [2022-02-04 00:43] Validation started with lhcb-run2-patches-dev4#957
- [2022-02-04 00:46] Validation started with lhcb-run2-patches#1093
- [2022-02-05 00:48] Validation started with lhcb-run2-patches-dev4#958
- [2022-02-05 00:48] Validation started with lhcb-run2-gaudi-head#274
- [2022-02-05 00:52] Validation started with lhcb-run2-patches#1094
- [2022-02-06 00:18] Validation started with lhcb-run2-patches-dev4#959
- [2022-02-07 00:36] Validation started with lhcb-run2-patches#1095
- [2022-02-07 00:36] Validation started with lhcb-run2-gaudi-head#275
- [2022-02-07 00:36] Validation started with lhcb-run2-patches-dev4#960
- [2022-02-08 02:41] Validation started with lhcb-run2-gaudi-head#276
- [2022-02-08 02:49] Validation started with lhcb-run2-patches-dev4#961
- [2022-02-08 02:50] Validation started with lhcb-run2-patches#1096
- [2022-02-09 00:41] Validation started with lhcb-run2-patches-dev4#962
- [2022-02-10 00:36] Validation started with lhcb-run2-gaudi-head#277
- [2022-02-10 00:39] Validation started with lhcb-run2-patches-dev4#963
- [2022-02-10 00:46] Validation started with lhcb-run2-patches#1097
- [2022-02-11 00:47] Validation started with lhcb-run2-gaudi-head#278
- [2022-02-11 00:48] Validation started with lhcb-run2-patches-dev4#964
- [2022-02-11 00:48] Validation started with lhcb-run2-patches#1098
- [2022-02-12 00:37] Validation started with lhcb-run2-patches-dev4#965
- [2022-02-13 00:22] Validation started with lhcb-run2-patches-dev4#966
- [2022-02-14 00:33] Validation started with lhcb-run2-patches-dev4#967
- [2022-02-15 00:36] Validation started with lhcb-run2-patches-dev4#968
Edited by Software for LHCb- Resolved by Patrick Koppenburg
Thanks for fixing this error.
For my understanding, what exactly prompts the crash? I would have naively thought that the
fill
call would write to a different variable name if, for some reason, the branch name changed it would just write to a different branch (obviously not the intended behaviour, but I'm just trying to understand what generates the fail). And, again out of interest, is theinitialize
function called for TupleTools every time a new file is opened?
mentioned in commit 422a6945