ENH: Option to store dataset creation/modification times with to_hdf.

[rickhg12hs]

Is your feature request related to a problem?

"I wish I could use pandas to" store a DataFrame with to_hdf and optionally include the datetime of dataset creation/modification.

Describe the solution you'd like

DataFrame.to_hdf should have an option that will store the datetime of dataset creation/modification.

[docstring addition]

with_datetime : bool, default False
    Stores `datetime` when dataset is written

API breaking implications

Shouldn't break anything.

Describe alternatives you've considered

A separate "metafile" that the user would need to write with dataset creation/modification datetimes.

Additional context

With the possibility of rewriting datasets, having the creation and modification times of the dataset in the HDF5 file would help to inform HDF5 file readers/users of updated datasets.

my_df.to_pdf(..., with_datetime=True, ...)
...
my_df_read = pd.read_hdf(store_file, key=key_value, mode="r")
print(my_df_read.ctime) # None or datetime
print(my_df_read.mtime) # None or datetime
...
my_df_read.info() # Would also show datetimes if present

[mroeschke]

Thanks for the suggestion.

My immediate reaction is that this seems out of scope for pandas as this problem doesn't seem entirely specific to HDF5 files (same can be said of CSV files for example), and alternate solutions like the one you provided, naming the file with timestamps, using the filesystem to keep track of time are valid solutions.

[rickhg12hs]

Understood.

Thanks for the consideration.