Planning November 2021

[SimonHeybrock]

Soon:

• Workshop on future/features of scipp plotting code
• Figure out how to handle "unaligned coords" (currently attrs) to free up the attrs property for Nexus/xarray-style attributes.
• Explore distributed computing (dask or alternatives)
  • Involve IDSs after initial internal discussion to get more concrete requirements/priorities
  • Do we need a more concrete outcome such as a script, or notebook template, or batch reduction GUI?
• API inconsistencies (also relating to Python array API specification)

Concrete tasks:

• Fitting (scipy wrapper) Curve fitting scipp#2325
  • May use same wrapping approach for peak finding Peak finding scipp#2326
• FFT (scipy wrapper)
  • Need complex dtype
  • or better smoothing from scipy.signal? Add scipp.signal providing butter and sosfiltfilt scipy wrappers scipp#2356
• Event-filtering toolkit (requires more investigation of requirements and consensus on desirable interfaces) Event filtering toolkit scippneutron#237
• bin-edge/histogram toolkit (requires more investigation of requirements and consensus on desirable interfaces) Histogram and bin-edge toolkit scippneutron#238
• (pulse-time) sliced file access Nexus
• Data stream listener, live visualization, live reduction
  • Get this to a point where it is easily deployed/launched
  • Live RAW data + reduced visualization
  • Configurable listener parameters, reduction parameters, reduction scipts
  • Config could be part of ESS instrument submodules

  python ess/launch_live_view.py --config ess/loki/live.conf

Uncategorized:

• Basic reduction workflows for many more instruments (all techniques)
• Naming of scipp software stack (how do we call scipp+scippneutron+ess)
  • What about names of scippneutron and ess?
  • What about potential new helper packages building on scipp or scippneutron, e.g., inelastic scattering normalization?
• Writing files for different analysis software packages
  • Do we need to be proactive about avoid code duplication, e.g., for various ASCII-based file formats
  • Can we or should we avoid doing this and focus on new (Nexus-based?) formats (with ECDC)
  • See, e.g., http://www.cansas.org/wgwiki/index.php/NXcanSAS
  • Support 1 standard per technique (but not multiple)?