You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I initially noted this issue in #87 but after a bit of investigation I realized it is affecting the basic kerchunk based HDF reader in main as well.
The problem
When a CF time encoded variable is written to an HDF file with compression (zlib in this case) and then opened via open_virtual_dataset with decode_times=True and serialized to kerchunk something is happening within the encoding process for "inlining" the CF time encoded variable that causes round trip comparisons to fail. This can be reproduced using this failing test 68f4426.
Cause?
I've poked around at this a bit but I'm not sure at which stage this problem is being introduced. Is something going awry with CFDateTimCoder with compressed data? Or is there a stray attr which is causing xarray to incorrectly decode the inlined byte data 🤔 . Hopefully someone with a bit more xarray-fu might have insights here.
The text was updated successfully, but these errors were encountered:
I initially noted this issue in #87 but after a bit of investigation I realized it is affecting the basic kerchunk based HDF reader in
main
as well.The problem
When a CF time encoded variable is written to an HDF file with compression (
zlib
in this case) and then opened viaopen_virtual_dataset
withdecode_times=True
and serialized tokerchunk
something is happening within the encoding process for "inlining" the CF time encoded variable that causes round trip comparisons to fail. This can be reproduced using this failing test 68f4426.Cause?
I've poked around at this a bit but I'm not sure at which stage this problem is being introduced. Is something going awry with
CFDateTimCoder
with compressed data? Or is there a strayattr
which is causingxarray
to incorrectly decode the inlined byte data 🤔 . Hopefully someone with a bit more xarray-fu might have insights here.The text was updated successfully, but these errors were encountered: