Currently, the only supported export option from a completed DeerLab FitResult is to pickle. Whilst pickling works well when everything is contained inside a single Python enviroment or computer it is not as good when transfering data between computer and Python versions. I am often finding that DeerLab fits completed on a spectrometer computer are not readable by my laptop.
My proposal would be to expand the FitResult class to contain an export method, as an additional option alongside pickling.
Advantages:
- This method would then export to some heirachial format that would be more easily shared and stored.
- This would also be much better for data archiving as instead of just saving a plot, the associated fit meta-data can also be kept.
- Most people in our community (for better or worse) use Matlab as their main data-processing and ploting method.
Disadvantages:
- Methods are not saved
- UQResult would not be saved (probably could still save the important info but would take more work)
Some potential data format options:
- HDF5 (Widly used in science but requires an extra package/dependency)
- Old-Matlab (Already included in the scipy package)
- New-Matlab (This is just a HDF5 wrapper)
- Numpy / .npz (Designed for multiple arrays, and only readable in Python)
What do other people think?
Currently, the only supported export option from a completed DeerLab FitResult is to pickle. Whilst pickling works well when everything is contained inside a single Python enviroment or computer it is not as good when transfering data between computer and Python versions. I am often finding that DeerLab fits completed on a spectrometer computer are not readable by my laptop.
My proposal would be to expand the FitResult class to contain an export method, as an additional option alongside pickling.
Advantages:
Disadvantages:
Some potential data format options:
What do other people think?