Iris 3.2.0 Deprecations

[bjlittle]

📰 Custom Issue

We require to consider and include any functional deprecations that we want to include for the forthcoming iris 3.2.0 release.

Existing Deprecations

• iris.util.as_compatible_shape. Deprecated in 3.0.0 (Removed iris.util.as_compatible_shape and now unused imports #4513)

Proposed New Deprecations for 3.2

• iris.util.approx_equal.
  Proposed by @rcomer, see comment below. Agreed
  (Deprecate iris.util.approx_equal #4514)
• iris.fileformats.abf.
  Proposed by @bjlittle. Agreed (remove - ask community if this is an issue)
  Also see https://github.com/SciTools/iris-abf and Add plugin support for loaders. #3657
  (Deprecate abf and dot #4515)
• iris.fileformats.dot.
  Proposed by @bjlittle. Agreed (remove - as above)
  (Deprecate abf and dot #4515)
• iris.experimental.raster.
  Proposed by @bjlittle. Agreed (remove - as above)
  (Deprecate experimental.raster #4525)
  (Add deprecation on routine usage. #4545)
• iris.experimental.regrid_conservative (and optional dependency esmpy).
  Proposed by @pp-mo. Agreed (remove - as above) Functionality available/equivalent in iris-esmf-regrid.
  (Deprecate regrid conservative #4551)
• iris.experimental.regrid
  Proposed by @pp-mo. Agreed (this will be moved)
  (Deprecate regrids #4548)

Investigate

• @rcomer proposal for intersection_of_cubes vs extract_overlapping, see comment below.
  (Deprecate intersection_of_cubes #4541)

Related

• See Tidy iris.experimental module #3515

Further

• We want to review and update the Deprecation Policy in our documentation (will also affect our page on Code Maintenance)
• Reference https://numpy.org/neps/nep-0023-backwards-compatibility.html

[rcomer]

How about iris.util.approx_equal? It appears to do the same job as math.isclose. Also I can't find any tests for it...

[rcomer]

I also think that iris.analysis.maths.intersection_of_cubes and CubeList.extract_overlapping do very similar things, though the latter is more flexible and better documented.