I’ve encountered that occasionally wrong values are stored for some branches for all the events in a file after taking a snapshot. It is reproducible in two different machines I’ve tried with different gcc versions. Making plot directly after Filter gives the correct answer, while making the plot byreading back from the snapshot’d filtered events gives wrong values. It doesn’t occur for all files. But when it does occur, all the events in a certain branch seems to get corrupted.
Hi,
We fixed several issues in RDF in the last weeks, could you please try to reproduce the problem with the current master branch? (Either by recompiling or taking it from CVMFS).
Hi @Suyong_Choi,
I see the problem too, it’s most probably a bug on our side.
This is now bug report ROOT-9860, you can follow the progress towards resolution there. As you see it’s marked with very high priority, so it should be resolved soon.
Thank you very much for the report and the simple reproducer!
Hi @Suyong_Choi, thanks to @Danilo the bug should now be fixed both on the current master branch and in the v6.16 branch (from which we will release v6.16 very soon).
It would be great if you could confirm that things work fine on your side now.
Thank you for the quick reply!
I will mark this as resolved, but do not hesitate to contact us should you encounter any further issue – or for any kind of feedback really.