Skip to content
jjpr-mit edited this page Apr 24, 2019 · 2 revisions

Things to choose:

  • bucket name
  • stimulus set name
  • image store unique name
  • image zip file name
  • assembly name
  • assembly store unique name
  • netCDF file name
  • names for metadata fields

Things that should always be the same:

  • The dimensions of the data array are ["neuroid", "presentation", "time_bin"].
  • No image-specific metadata is on the data array, only "image_id"; all image-level and higher metadata (like object and category) is stored in lookup.db.

Steps:

  • Make .nc file of DataArray in correct format
    • rearrange and stack as neuroid, presentation, time_bin
    • make .nc file
  • Make image zip
  • make StimulusSet lookup meta
    • check list of EAVs
  • make DataAssembly lookup meta
  • test locally
    • put .nc in place
    • put zip in place
    • do get_assembly
  • make bucket
  • put image zip in bucket
  • put .nc in bucket
  • check in and push the packaging script
  • test by setting aside and doing get stimulus set and get assembly
  • git branch
  • commit lookup
  • add unit test and commit
  • do pull request
  • switch local branch back to master
Clone this wiki locally