root_pandas
root_pandas copied to clipboard
scikit-hep
Flattening two branches produces only one `__array_index`
I have a tree with DecayTreeFitter and when I retrieve two array variables and use the flatten=True option with the read_root function, the resulting dataframe only contains a single __array_index column, so it's not possible to tell which index belongs to which column.
Side note: Addressing #28 would fix this, I may have a look into implementing it.
From examining the tests, it looks like using flatten=True is deprecated (please correct me if I'm wrong) and should produce a warning.
For some reason the warning does not come up for me.
Only one __array_index is produced at the moment, because we only support flattening arrays with the same length, in which case __array_index is valid for both variables at the same time.
It's strange that it doesn't give you a warning. Pinging @KonstantinSchubert
I should also update the documentation to reflect the changed API.
After looking at the code, it looks like the warning is not in the current pip version. Should be fixed as soon as the new version is pushed :)
Re: __array index flattening, I think docs is enough to fix this problem for now, but #28 would be a much nicer way to fix it imo.
@nasfarley88: The functionality suggested in #28 doesn't really overlap with the flatten kwarg, because flatten is mainly designed to work with variable-length arrays, which #28 can't handle.
Btw, in the current pip version, we're still using flatten=True/False, so the documentation is currently up to date for that.
But we should be more clear about what happens when you are flattening multiple arrays.
I'll also make a new release soon, because the list argument to flatten fixes several important issues, like the fact that flatten=True/False is too greedy and tries to flatten all arrays, regardless whether they are compatible in length, which leads to an error.
As explicitly written in the README since a while, root_pandas, and root_numpy on which it depends, has been deprecated and effectively unmaintained for quite a while. We decided to close anthing outstanding as "won't do" and archive the package at this point.
