Breaking change in ROOT files created with ROOT6

[aleberti]

Description of the problem can be found in https://root-forum.cern.ch/t/ttime-saved-to-a-ttree-in-root5-and-root6/54031 and scikit-hep/uproot5#859 , found by David Green while using uproot within pybkgmodel (https://github.com/cta-observatory/pybkgmodel).

Apparently a bug in ROOT6, where old I/O is used for certain (historical) classes. Reported also to uproot developers. Affects ROOT6 versions up to 6.28 (current one at the time of the opening of this issue)

Does this affect us? Yes and no.

• no because OSA still uses ROOT5 (and there are no plans to upgrade it to ROOT6, the machine has an old operative system where currently ROOT6 cannot be installed), so all files up to superstar will have the MTime information as it is now, so perfectly readable by uproot
• no because there is a temporary fix (see again Getting information from TTime fMilliSec member scikit-hep/uproot5#859)
• yes in the case we will have re-calibration of some joint data, which is done at PIC using ROOT6. At PIC for the moment we are stuck with ROOT6.24.08 (due to old OS present there), where this issue is present (maybe we could patch it up when ROOT developers fix the issue). So if re-calibration is performed at PIC, the calibrated files will not be readable by ctapipe_io_magic as it is now.

Anyway, ROOT and uproot developers have been notified, and now also in contact for a related but somewhat different issue.

[aleberti]

Will be solved in upcoming ROOT release v6.30, see root-project/root#12499

[aleberti]

As an update, as anticipated, this now should be solved with ROOT v6.30.02, recently released: https://root.cern/doc/v630/release-notes.html#release-6.3002

[aleberti]

Just tried to produce e.g. some Melibea files with ROOT 6.30.02 and reading with uproot 5 does not show the issue anymore.