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
While R is great, and the bindings that @benfasoli wrote for STILT are incredible, for users, like myself, whose language of choice is something other than R, outputting the trajectories to rds format is quite inconvenient. Sure, we could use R conversion libraries like rpy2, but it would be much more convenient if the trajectories could be written to a cross-platform, portable format.
Some ideas include:
HDF5 - offers compression to columnar data with the added support of attributes
netCDF4 - standard in geoscience, less compressed?
The text was updated successfully, but these errors were encountered:
HDF5 and NetCDF4 are probably already installed since the footprints are NetCDF4 but they aren't very ergonomic for users and usually have clunky install stories because the language bindings are links to the system-level HDF5/NetCDF4 C-lib installs.
Take a look at parquet files for a performant column storage format that works cross-platform without many of the same drawbacks 👍
I have implemented parquet as an option and added documentation. I left h5 as an option, but didnt document in the markdown docs. I also added the option and documentation to disable writing trajectories with ''
While R is great, and the bindings that @benfasoli wrote for STILT are incredible, for users, like myself, whose language of choice is something other than R, outputting the trajectories to
rds
format is quite inconvenient. Sure, we could use R conversion libraries likerpy2
, but it would be much more convenient if the trajectories could be written to a cross-platform, portable format.Some ideas include:
HDF5
- offers compression to columnar data with the added support of attributesnetCDF4
- standard in geoscience, less compressed?The text was updated successfully, but these errors were encountered: