Generalise support for netcdf load+save of "special" attributes #6566
+1,055
−92
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
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #6566 +/- ##
==========================================
- Coverage 89.94% 89.93% -0.02%
==========================================
Files 90 91 +1
Lines 24302 24385 +83
Branches 4546 4554 +8
==========================================
+ Hits 21859 21930 +71
- Misses 1674 1686 +12
Partials 769 769 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
pp-mo
added
the
benchmark_this
⏱️ Performance Benchmark Report: 71c7bc4Performance shiftsFull benchmark resultsGenerated by GHA run |
pp-mo
force-pushed
the
nc_attribute_handlers
branch
from
ad34c83 to
dc0c441
Compare
…d/save code, not handlers.
…hout split-attrs.
pp-mo
force-pushed
the
nc_attribute_handlers
branch
from
afa11e5 to
9d90853
Compare
Review Notes : changes to considerThe code changes here provoked review of some older behaviour decisions. Here's a summary, which may not be entirely complete: overall
STASH
ukmo__process_flags
GRIB_PARAM
|
Uncertainty remaining
|
pp-mo
force-pushed
the
nc_attribute_handlers
branch
from
78d96e8 to
7226b36
Compare
Update:
I'm now thinking that, without any further changes here, I can just import the tests which are here skipped due to iris-grib not being installed, and run them in the iris-grib repo instead. Does that sound like a good plan ? |
bjlittle
requested changes
Jul 22, 2025
There was a problem hiding this comment.
@pp-mo Looks good.
You've had to grapple with some serious sketchy historical code that's evolved over time. Unifying it should make it more understandable and maintainable.
I can't see any obvious behaviour changes that will upset users, so I'm happy to roll with what you've suggested and we can see whether there are any user edge cases that may require a patch or rebuffing.
Thanks for adding the additional test coverage 🤩💯
Only a few minors to service 👍
| """ | ||
|
|
||
| from abc import ABCMeta, abstractmethod | ||
| from typing import Any, Dict, List, Tuple |
There was a problem hiding this comment.
@pp-mo As from >=py39 (PEP585) you now no longer need to import Dict, List or Tuple. It's safe to use the built-in equivalents i.e., dict, list and tuple 👍
| NetcdfIdentifyingNames: List[str] = [] | ||
|
|
||
| @abstractmethod | ||
| def encode_object(self, content) -> Tuple[str, str]: |
There was a problem hiding this comment.
@pp-mo Do you we have a type annotation for content?
There are other various missing annotations below ...
Not too sure how strict you want to be on type annotations just yet in iris ... the coverage and adoption is patchy at best.
I'd err on the side of either typing everything new or not at all. Your call.
| pass | ||
|
|
||
| @abstractmethod | ||
| def decode_attribute(self, attr_name: str, attr_value) -> Any: |
There was a problem hiding this comment.
@pp-mo attr_name is never used in any the subclass methods, so why is it part of the abstract method?
Is there something I've misunderstood?
| ) | ||
| warnings.warn(msg, category=_WarnComboLoadIgnoring) | ||
|
|
||
| if len(matches) > 0: |
There was a problem hiding this comment.
Suggested change
| if len(matches) > 0: | |
| if len(matches) == 1: |
| var = engine.cf_var | ||
| for handler in ATTRIBUTE_HANDLERS.values(): | ||
| # Each handler can have several match names, but ideally only 0 or 1 appears ! | ||
| iris_name = handler.IrisIdentifyingName |
There was a problem hiding this comment.
@pp-mo Super minor, but you could offer a read-only property, for example, on the handler that allows the user of the instance to get the iris or netcdf name without having to know the name of the class property that they're stored in e.g., handler.iris_name and handler.netcdf_name
Happy for you not to do this, just an observation/suggestion.
github-project-automation
bot
moved this from 🚀 In Progress
to 👀 In Review
in 🦋 Iris 3.13.0
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.
Closes SciTools/iris-grib#596 and SciTools/iris-grib#674
Ties together the handling of STASH, GRIB_PARAM and "ukmo__process_flags" attributes,
which all require special handling on netcdf load+save.
By design, should preserve existing behaviour.
But I have tried to generalise the handling,
( and also rationalise behaviour a wee bit, at least for "ukmo__process_flags" -- though that one is quite obscure ).
TODO:
@staticmethod+ useselfmore(but first, just let's see if it breaks anything I haven't already tried)