Ready for 06 by jo-mueller · Pull Request #605 · ome/ome-zarr-py

jo-mueller · 2026-07-24T08:37:02Z

@will-moore maybe one more thing before shipping the latest release. This PR

replaces 0.6.dev4 by 0.6 in the written and read versions
adds some more checks. Prior to this PR, some of the version check clauses missed the check for 0.6 so some of the additional metadata (i.e., omero and image-label) would never be ingested in case a version 0.6x were encountered.
Adds 0.6 in the tests more broadly

Rebased on #609

for more information, see https://pre-commit.ci

codecov · 2026-07-24T08:56:57Z

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 86.51%. Comparing base (e65666e) to head (89ae7f8).

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #605      +/-   ##
==========================================
+ Coverage   86.46%   86.51%   +0.04%     
==========================================
  Files          16       16              
  Lines        2365     2366       +1     
==========================================
+ Hits         2045     2047       +2     
+ Misses        320      319       -1

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

will-moore · 2026-07-29T11:04:23Z

-            # Validate with ome-zarr-models-py: only supports v0.4
-            Models04Image.from_zarr(out)
-        elif version.version == "0.5":
-            Models05Image.from_zarr(out)


Do we need to remove these checks?

We don't have to, but since the metadata is essentially ingested with ome-zarr-models-py, we are kind of already validating on-the-fly

will-moore · 2026-07-29T11:56:32Z

In order that the write_image(data) method creates v0.6 data, we need a FormatV06 class to be the CurrentFormat.
If we also add this to the tests, that looks like...

diff --git a/ome_zarr/format.py b/ome_zarr/format.py
index 3b2c33e..583fe18 100644
--- a/ome_zarr/format.py
+++ b/ome_zarr/format.py
@@ -384,4 +384,14 @@ class FormatV05(FormatV04):
         return {"name": "default", "separator": "/"}
 
 
-CurrentFormat = FormatV05
+class FormatV06(FormatV05):
+    """
+    Changelog: added FormatV06 (May 2025): writing not supported yet
+    """
+
+    @property
+    def version(self) -> str:
+        return "0.6"
+
+
+CurrentFormat = FormatV06
diff --git a/tests/test_writer.py b/tests/test_writer.py
index 836b746..4c7bb58 100644
--- a/tests/test_writer.py
+++ b/tests/test_writer.py
@@ -31,6 +31,7 @@ from ome_zarr.format import (
     FormatV03,
     FormatV04,
     FormatV05,
+    FormatV06,
     format_from_version,
 )
 from ome_zarr.scale import _build_pyramid
@@ -58,6 +59,7 @@ TRANSFORMATIONS = [
 FORMAT_VERSIONS = [
     pytest.param(FormatV04, id="V04"),
     pytest.param(FormatV05, id="V05"),
+    pytest.param(FormatV06, id="V06"),
 ]

This creates a few failures which I haven't looked into yet....

jo-mueller · 2026-07-29T12:10:59Z

In order that the write_image(data) method creates v0.6 data, we need a FormatV06 class to be the CurrentFormat.

right...I was hoping to remove the Format class at some point (all of its functionality is now present elsewhere). But for the sake of going forward with 0.6, I could add the FormatV06 class here and mark it for later removal?

for more information, see https://pre-commit.ci

jo-mueller · 2026-07-30T12:40:52Z

This creates a few failures which I haven't looked into yet....

oh boy. Jep, quite a few. I think this requires a bit of refactoring until all functions that are writing data somehow are 0.6 compliant one way or another 🙈

jo-mueller · 2026-07-31T09:10:52Z

@will-moore ok, this is turning out super-hard, because there are still more than writer paths (class-based, write_image/write_labels, write_multiscale/write_multiscale_labels), all of which would somehow need to be updated to write 0.6, where only the first one (class-based) is really fit for it at this point. And I kind of don't want to update code that we want to deprecate in the long run anyway.

So my suggestion:

I add an option in the OMEZarrMultiscale class that it can be constructed either
- from a single OMEZarrImage and the pyramid is built under the hood. The write_image and write_labels already use that under the hood.
- from a list of OMEZarrImages: This would be new and would be equivalent to writing a pre-computed pyramid. This could then be used in the write_multiscale and write_multiscale_labels function

This way, we would get a few props

The testing data for all these three paths (class based, write_image, write_multiscales) would effectively take the same path through the writer and could be simplified
Compliance with 0.6 would need to sit only at one place (inside the OMEZarrMultiscale class)
There are cases where creating OMEZarrMultiscale from pre-computed pyramids is very desirable, for instance for (down-)loading data. In this case, the resolution levels are effectively precomputed and one would decidedly not want to create the downscaled copies from a potentially large image. If we were to do this, we would download a big image and downscale instead of downloading a small one to begin with.

In the meantime, I would revert the changes here to dc07c55 (before the introduction of the FormatV06 class) so we can ship the pre-release.

This reverts commit e0a03cd.

This reverts commit 75b9402.

This reverts commit b12fb12.

jo-mueller · 2026-07-31T09:48:30Z

Oh no. And tests are failing with something compression related 💀

will-moore · 2026-07-31T10:27:36Z

Just trying to see what changed in the failing tests, looking at installed versions. Since the tests passed previously, zarr 3.3.0 has been released, and it's only installed on python 3.12 and 3.13 (failing tests). Older zarr is used on the py 3.11 tests which are passing. But I don't see anything obvious in https://github.com/zarr-developers/zarr-python/releases/tag/v3.3.0

will-moore · 2026-07-31T10:47:44Z

Your suggestion sounds good.

Re: deprecating Format classes - I see that image.to_ome_zarr() creates FormatV05() when writing v0.6 because the underlying _write_pyramid_to_zarr() method expects a Format, so trying to untangle these is a going to be a bit of work.

jo-mueller · 2026-07-31T11:18:59Z

so trying to untangle these is a going to be a bit of work.

Jep. My approach is to remove the metadata writing entirely from the _write_pyramid_to_zarr entirely, because the written metadata there is overwritten by the ozmp model_dump later anyway. That will make the write_multiscale_metadata obsolete.

_write_pyramid_to_zarr will only need to check the format to see whether it needs to go for zarr v2 or zarr v3. PR coming soon!

jo-mueller · 2026-07-31T13:31:00Z

For now, I'll pin zarr-python<3.3.0

jo-mueller · 2026-07-31T14:20:22Z

Ok, this is looking good. Pinging @d-v-b here - maybe he knows whether something has changed on the zarr side that could have breaking effects? For reference, tests here started failing in a9da128 after the release of zarr-python 3.3.0.

d-v-b · 2026-07-31T14:25:12Z

can you link me to a log with the test failures in it?

jo-mueller · 2026-07-31T14:37:29Z

https://github.com/ome/ome-zarr-py/actions/runs/30620108525/job/91124450262

d-v-b · 2026-07-31T14:42:10Z

got it, fix incoming

zarr 3.3.0 changed codec configuration attributes such as `BloscCodec.cname` and `BytesCodec.endian` from `Enum` members to plain strings, breaking the `.name` accesses in the label writer storage options tests (12 failures). Add a `_codec_value` helper that normalizes both representations, so the tests pass against zarr 3.2 and 3.3 alike. Test-only change; the library code never accessed these attributes. Assisted-by: ClaudeCode:claude-opus-5

This reverts commit dcd3858.

jo-mueller · 2026-07-31T16:52:59Z

Ok, will rebase on #609 or wait until that is in.

jo-mueller added 4 commits July 24, 2026 10:33

feat: write 0.6, not 0.6.dev4

2f972c3

feat write 0.6

129c351

feat: add missing version checks in additional metadata

ee3bc31

tests: add 0.6 to tested versions in writer

37149fb

jo-mueller added the bug Something isn't working label Jul 24, 2026

pre-commit-ci Bot and others added 2 commits July 24, 2026 08:37

[pre-commit.ci] auto fixes from pre-commit.com hooks

40c7a70

for more information, see https://pre-commit.ci

tests: pull out scale transforms correctly irrespective of version

9c7e4f0

will-moore reviewed Jul 29, 2026

View reviewed changes

Comment thread ome_zarr/classes/image.py Outdated

jo-mueller and others added 4 commits July 30, 2026 13:21

feat: default to 0.6 for writing

dc07c55

feat: add formatV06 class

b12fb12

tests: add checks for what version is encountered

75b9402

[pre-commit.ci] auto fixes from pre-commit.com hooks

e0a03cd

for more information, see https://pre-commit.ci

jo-mueller added 3 commits July 31, 2026 11:29

Revert "[pre-commit.ci] auto fixes from pre-commit.com hooks"

934096c

This reverts commit e0a03cd.

Revert "tests: add checks for what version is encountered"

62ecec6

This reverts commit 75b9402.

Revert "feat: add formatV06 class"

a9da128

This reverts commit b12fb12.

jo-mueller mentioned this pull request Jul 31, 2026

feat: Accept pyramids #608

Open

will-moore reviewed Jul 31, 2026

View reviewed changes

Comment thread pyproject.toml Outdated

deps: pin to zarr-python <3.3.0

dcd3858

jo-mueller force-pushed the ready-for-06 branch from 5375584 to dcd3858 Compare July 31, 2026 14:12

d-v-b and others added 2 commits July 31, 2026 16:57

Revert "deps: pin to zarr-python <3.3.0"

90be359

This reverts commit dcd3858.

Merge branch 'claude/zarr-codec-enum-compat-d9bf40' into ready-for-06

89ae7f8

jo-mueller mentioned this pull request Aug 4, 2026

Ome zarr scene #612

Open

Uh oh!

Conversation

jo-mueller commented Jul 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

codecov Bot commented Jul 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

will-moore Jul 29, 2026

Choose a reason for hiding this comment

Uh oh!

jo-mueller Jul 29, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

will-moore commented Jul 29, 2026

Uh oh!

jo-mueller commented Jul 29, 2026

Uh oh!

jo-mueller commented Jul 30, 2026

Uh oh!

jo-mueller commented Jul 31, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jo-mueller commented Jul 31, 2026

Uh oh!

will-moore commented Jul 31, 2026

Uh oh!

will-moore commented Jul 31, 2026

Uh oh!

jo-mueller commented Jul 31, 2026

Uh oh!

jo-mueller commented Jul 31, 2026

Uh oh!

Uh oh!

jo-mueller commented Jul 31, 2026

Uh oh!

d-v-b commented Jul 31, 2026

Uh oh!

jo-mueller commented Jul 31, 2026

Uh oh!

d-v-b commented Jul 31, 2026

Uh oh!

jo-mueller commented Jul 31, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

jo-mueller commented Jul 24, 2026 •

edited

Loading

codecov Bot commented Jul 24, 2026 •

edited

Loading

jo-mueller commented Jul 31, 2026 •

edited

Loading