Correctly generate NWB2 files with all previously existing fields

[alessandrofelder]

A list of things we should check to ensure a consistent file generation for NWB2 files, based on comparing output NWB2 signature and existing signature files (from NWB1) at commit 5898a0f for sample_pointing_videos_161215_15_34_21 and sample_miniscan_fred_170322_14_06_43

related to #8

Missing in new NWB output

• /general/optophysiology/cycle_time

• /general/optophysiology/frame_size

• /general/optophysiology/imaging_mode

• /general/optophysiology/zplane_pockels

• /general/specifications/e-lab-view.py

• /general/specifications/e-pixeltimes.py

• /general/specifications/nwb-core.py

• epoch data stored differently -> link to corresponding acquisition missing

• value for /stimulus/presentation/air_puff/

• descriptions for

  • /general/devices/AOL_microscope
  • /general/devices/eyecam
  • /general/devices/whiskerscam
  • /general/devices/air_puff_device
  • /general/devices/mouse_wheel_device

• labview header

Newly there in NWB2

• all timestamps entries have an @interval field with dtype=int32, shape=(), val=1
• some info on epochs and trials
• one line on /timestamps_reference_time
• /acquisition/ROI has extra shape dimension for 0027,0029 and 0030 (all both Red and Green)

Different in NWB2

• all timestamps have different hash values apart from a few cases (ROI 0027,0029 and 0030) where the hash value is and was [0]
• Zstack0001-0049/imaging_rate and ROI_nnnn_col/scan_line_rate values (sometimes) differ (only by <1e-6)
• very different organisation of /processing/Acquired_ROIs/ImageSegmentation/
• For EyeCam and WhiskersCam:
  • frame rate in new output likely wrong and inverse of previous value
  • external file has now non-ignored type and value
• /acquisition/Eyecam/@pmt_gain and scan_line_rate have dtype=float64
• /acquisition/ROI/data now has dtype=int8 instead of float32 (0001-0010) or uint16(0027,0029 and 0030) (all both Red and Green)
• /acquisition/ROI/data now has different hash values

[jonc125]

See also #6.

[ageorgou]

All handled in #13, I think?

[alessandrofelder]

I think so, too