6

When using pandas.to_hdf function to save data to a HDF5 file, I'm getting the following warning:

C:\{my path to conda environment}\lib\site-packages\tables\file.py:426: 
UserWarning: a closed node found in the registry: ``/{my object key}/meta/{my column name}/meta/_i_table``
  warnings.warn("a closed node found in the registry: "

I wasn't able to google what this warning could mean and looking into the relevant PyTables docs didn't help.

Does anybody have an idea?

Details:

  • Python 3.8.2, PyTables 3.6.1, pandas 1.0.3
  • The saved data seem to be alright.
  • I'm calling the pandas.to_hdf function as follows: to_hdf(filename, key, mode = mode, complib='blosc', format = 'table', append = True), where mode is either 'w' or 'a' (the warning is raised in both cases).
PGlivi
  • 996
  • 9
  • 12

2 Answers2

0

For all its worth i managed to overcome the warning by installing tables 3.6.1 from: https://www.lfd.uci.edu/~gohlke/pythonlibs/#pytables using the +gpl version for python 3.9.

Never the less i do not know the details of this choice and i cannot justify why it is working.

Kostas Markakis
  • 143
  • 2
  • 11
0

I observed the same issue. For me, this happens (from time to time, I do not see the logic) for category columns in the DataFrame.

Solution:

  • transform category columns to object
  • run to_hdf(...)
PanchoVarallo
  • 145
  • 1
  • 9