With the .gz extension, df.to_pickle and pd.read_pickle can automatically decide to use Gzip for compression/expansion without explicitly specifying the compression argument.
Thanks to the above, we can remove the compression argument. This also allows users to use their preferred compression formats like .bz2.
It would be also nicer to use .pkl (or .pickle) rather than pckl because
Much more usage of .pkl.gz can be found on web in general than .pckl.gz. So people can more easily notice that this is in the pickle format.
For already made pckl.gzip files, we can simply rename them.
Other changes
From an example in docs/pacemaker/quickstart.md, I removed the protocol argument from pd.read_pickle (this argument exists for df.to_pickle but not for pd.read_pickle).
Based on #30
Summary
This suggests using the
.pkl.gz
extension rather than the present.pckl.gzip
extension for dataset files.Motivation
There are several advantages to do so:
.gz
is the standard extension for Gzip files (see https://www.gnu.org/software/gzip/manual/gzip.html and https://en.wikipedia.org/wiki/Gzip)..gz
extension,df.to_pickle
andpd.read_pickle
can automatically decide to use Gzip for compression/expansion without explicitly specifying the compression argument..bz2
.It would be also nicer to use
.pkl
(or.pickle
) rather thanpckl
because.pkl
extension in their examples (seedf.to_pickle
andpd.read_pickle
)..pkl.gz
can be found on web in general than.pckl.gz
. So people can more easily notice that this is in the pickle format.For already made
pckl.gzip
files, we can simply rename them.Other changes
From an example in
docs/pacemaker/quickstart.md
, I removed the protocol argument frompd.read_pickle
(this argument exists fordf.to_pickle
but not forpd.read_pickle
).