Skip to content

Added scan connector with blissdata

Valentin Valls requested to merge fix-bliss-data into main

This PR rework some stuffs then implement a scan connector with Blissdata.

Probably closes #117 (closed)

  • Use pytest tmp dir for nexus writer files (closes #119 (closed))
  • Allow to close components in order to clean up env (mandatory for some resources)
  • Setup a minimalistic app to only test independent components
  • Implement blissdata (which fallback to hdf5 itself)
  • Allow to specify bliss/blissdata as source, if bliss>=2.0 -> blissdata is used

This could be merge, but other improvement have to be done in blissdata itself.

Plus few warn

  • I haven't tested with the real application
  • There is a limitation in blissdata scan search which crop the result to 2000 scans (in order to not explode the memory)
    • Plus, during a proposal, metadata for eachs scans stays in the DB, at some places you could have 2000 scans at the end of the week
    • Maybe blissdata API have to be improved to
      • At least allow to take a look at the last n scans
    • Maybe blissdata/daiquiri API/code have to be improved to
      • At best allow to page the scan result without reaching the whole scan keys
  • BLissdata is supposed to only contain scan from the actual proposal, but actually it is not the case
    • We can filter the scans by proposal, but it is done in this version
  • Instead of scanid we should use scan_key in mimosa at some point
Edited by Valentin Valls

Merge request reports