Improve consistency of default engine and return memoryview instead of bytes from to_netcdf() #10656
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This PR introduces a bug fix and a breaking changes: 1. The default backend ``engine`` used by `Dataset.to_netcdf` and `DataTree.to_netcdf` is now chosen consistently with `open_dataset` and `open_datatree`, using whichever netCDF libraries are available and preferring netCDF4 to h5netcdf to scipy. Previously, `DataTree.to_netcdf` was hard-coded to use h5netcdf. 2. The return value of `Dataset.to_netcdf` without ``path`` is now a ``memoryview`` object instead of ``bytes``. This removes an unnecessary memory copy and ensures consistency when using either ``engine="scipy"`` or ``engine="h5netcdf"``. Fixes pydata#10654
github-actions
bot
added
topic-backends
topic-DataTree
shoyer
commented
Aug 19, 2025
| @dataclass | ||
| class BytesIOProxy(Generic[BytesOrMemory]): | ||
| """Proxy object for a write that returns either bytes or a memoryview.""" | ||
| class BytesIOProxy: |
There was a problem hiding this comment.
Note: I'm keeping around BytesIOProxy because we'll need it for #10624
shoyer
changed the title
OriolAbril
reviewed
Aug 25, 2025
There was a problem hiding this comment.
I didn't realize the PR fixing the issue was open already. Thanks for opening it. I have added a comment regarding adding this as a breaking change. I realized about the issue because code that has run successfully in CI since the introduction of DataTree in xarray stopped working with 2025.8.0.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR introduces two breaking changes:
engineused byDataset.to_netcdfandDataTree.to_netcdfis now chosen consistently withopen_datasetandopen_datatree, using whichever netCDF libraries are available and valid, and preferring netCDF4 to h5netcdf to scipy. Previously,DataTree.to_netcdfwas hard-coded to use scipy for writing to file-like objects or bytes, andDataTree.to_netcdfwas hard-coded to use h5netcdf.Dataset.to_netcdfwithoutpathis now amemoryviewobject instead ofbytes. This removes an unnecessary memory copy and ensures consistency when using eitherengine="scipy"orengine="h5netcdf".It also includes a minor bug-fix, raising an error when returning a memoryview with
compute=FalseFixes #10654
whats-new.rst