.. _roadmap:
Roadmap
=======
Overview
--------
The overall plan for the coming releases is to:
- Consolidate the existing generic functionality.
- Add missing generic functionality to provide a more complex toolbox.
- Weed out usability hurdles, to ensure the library's interface is easy to use.
At this point we do not have a fixed release cycle.
Instead we publish new releases when major feature additions or breaking changes have gone into Scipp.
Upcoming milestones
-------------------
We are currently consolidating and simplifying Scipp's core data structures.
In particular:
- ``scipp.Dataset`` will be restricted to items of matching dimensionality (``scipp.DataGroup`` can be used for other applications).
- Support for attributes will be removed from ``scipp.DataArray`` due to conceptual consistency issues.
Future direction and plans
--------------------------
We have a number of improvements in mind which we would like to consider and potentially introduce to Scipp.
At this point we have not prioritized, scheduled, or funded any of these:
- `dask `_ support.
- Support for `numpy's array-function API `_, for better interoperability.
- Experiment with moving more high-level functionality from C++ to Python.
Could we wrap NumPy (or similar) arrays, or is the performance penalty too high?
- Better support for structured dtypes and spatial dtypes.