Skip to content

Dangling references to other objects in zarr tests #17

Closed

Description

@DragaDoncila

A good chunk of the current zarr cases for v0.6 contain references to objects that aren't there e.g. here where we reference a multiscale at 0/ but we don't have anything there.

I think we want to add these objects right? @clbarnes

Activity

  1. clbarnes commented on Oct 1, 2026

    @clbarnes
    Collaborator
  2. clbarnes commented on Oct 1, 2026

    @clbarnes
    Collaborator

    I would suggest updating the validate_zarr/*/transforms/* tests to use a Scene rather than a Multiscale because multiscales require adding at least one zarr array, where a Scene doesn't strictly need to actually contain any multiscales so you can get away with just one file.

  3. kevinyamauchi commented on Oct 12, 2026

    @kevinyamauchi
    Collaborator

    2. Todo list here: https://docs.google.com/spreadsheets/d/1eHHbY0B1_IqIBXAblQFSjY4tRQxd4Mpw9AkFqgss6_M/edit?usp=sharing
    Os for todo, Xs for done.

    Hey @DragaDoncila and @clbarnes ! I have capacity to ramp up on fixing these tests this week. However, I'm realizing that I don't quite understand what we need to fix 😬 The OP in this issue seems to be about dangling references in the validate_zarr tests, but the spreadsheet references the parse_attributes tests. Can somebody give me a bit of context/help getting started? Happy to jump on a call if that's easier

  4. clbarnes commented on Oct 12, 2026

    @clbarnes
    Collaborator

    At a minimum, every parse_attributes test should have a validate_zarr test, because some tools will only want to operate on whole zarr hierarchies (e.g. in languages with privacy, if the metadata structures are an internal implementation detail - useful when abstracting over multiple versions). So I've listed all the parse_attributes tests to stand in as a list for which validate_zarr tests should exist.

  5. kevinyamauchi commented on Oct 29, 2026

    @kevinyamauchi
    Collaborator

    @clbarnes , I think that we can close this now that #48 is merged. What do you think?

  6. kevinyamauchi commented on Oct 29, 2026

    @kevinyamauchi
    Collaborator

    Everything on the spreadsheet is ticked off, so closing this issue. Thanks everyone!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions