v3.1.x

[bjlittle]

The release manager for v3.1.x is @pp-mo

The release feature branch is v3.1.x

Minor Release	Status	Milestone	Project	Date	RTD	Release	conda-forge	PyPI
v3.1a0†	Completed	v3.1.0	Iris v3.1.0	03/05/2021	-	-	-	-
v3.1.0rc0	Completed	v3.1.0	Iris v3.1.0	17/08/2021	Whatsnew	v3.1.0rc0	Available	-
v3.1.0	Completed	v3.1.0	Iris v3.1.0	17/09/2021	Whatsnew	v3.1.0	Available	Available

Notes:

• † - v3.1a0 is an internal engineering tag.

[pp-mo]

Final planning for content now happening (at last) !

Overall plan:

• we are still intending to get an RC out for the end of August 2021 : working on this now
• we simply don't have time now to include the prototype unstructured(UGRID) support in the mesh-data-model feature-branch.
  • which is a shame as this could be disruptive and we want to get it "out there", but for now it must wait.
  • .. but equally, mesh-data-model is still not quite ready, and this simplifies the 3.1 task massively

I will shortly be adding notes here on proposed content, including :

• tasks;
• outstanding PRs;
• whatsnew additions and missing content

[pp-mo]

PRs to be included

The list should now be complete on the 3.1.0 release project
However, not everything there is 100% ready to go, and we might still have to "pull" some of them from the release.

These are all the possibles + their status, from the project columns :
Waiting for review

• Add barbs plot #3710
  • possibly all-done. Pinged contributor. Awaiting response + review signoff/merge from dev
• Cubeprinter adopt #4206
  • Needs reviewing. Also possibly a whatsnew entry (@pp-mo working on that)

In progress

• Docs: Enable spell check #3882
  • not quite ready. Pinged contributor. Not vital, but we will complete this + include it if we can
• Cube arithmetic array type handling #3790
  • status slightly unclear, may in fact be ready for review. Pinged contributor
• Added AtmosphereSigmaFactory #4052
  • needing at least rebase + review. Pinged contributor
• Simple support for selective netcdf loading. #4176
  • still in draft, and rather wanted by ESMValTool people.
  • this one is on me (@pp-mo) : status: will fix in ~2weeks when back from holiday, or someone else could take it on ?

Backlog
There are the 3 outstanding tasks relating to improvements/fixes to the notebook Cube formatting Cube._repr_html()
All from @DPeterK , who I have now pinged

• Catch failed html_repr for cube #3376
• Cube HTML repr CSS improvements #3380
• Add dask array html repr to cube html repr #3383
• NOTE: I moved all these to 'TODO', below the MVP-divider, for two reasons:
  1. we suspect there may not be time to get these in (they have been stale for a while)
  2. they may be affected by Cubeprinter adopt #4206, or at least better re-built around Add abstract cube summary #3987

[rcomer]

I just added #4263 to the project, as it is a fix for a problem I introduced since v3.0 😬

[pp-mo]

Notes on additional Whatsnew content

I checked out all the commits from v3.0.0 to current 3.0.x head, looking for any significant changes or contributions lacking a whatsnew entry.
Which results in this list ...

• Corrected plot_anomaly_log_colouring for new Matplotlib linscale rules. #4115 : worth a mention, in the "FIXES" section?
• Gallery: don't support mpl v1.2 #3941 : in "Internal"
• Widen cube printout for long ancil or cell-measure names. #4124 : this does appear in v3.0.4, but quoting two different PRs against 3.0.x : Wide cubestr fix 3v0vx v2 #4233 and (More of) Wide cubestr fix 3v0vx v2 #4238
• Replace pyke nopyke #4198 : should really have an "Internal" section entry
  • propose text something like "removed the dependency on Pyke for netcdf loading, replaced with pure Python."
• Add abstract cube summary #3987 : should be mentioned in entry for Cubeprinter adopt #4206, assuming that gets in
• Added AtmosphereSigmaFactory #4052 missing whatsnew

There have also been a lot of small improvements to documentation content + presentation, mostly from @tkknight and @bjlittle . It might be worth saying something about this, but I'm strugggling to summarise the changes.

All best dealt with in final stages of release definition, I think.

[rcomer]

For #3941, I figured it was so minor it didn’t warrant a whatsnew entry. If you disagree, it could just be added to point 1 under documentation (other gallery updates).

https://github.com/SciTools/iris/blob/eca4fa16ece4e1a001420f939a4b6062824a1981/docs/src/whatsnew/latest.rst#-documentation

[pp-mo]

Thanks @rcomer . You're right -- no-one is still waiting to be told that we don't support mpl 1.x !

[pp-mo]

I haven't seen any suggestions to modify the 3.1 release.
However, from discussion I am thinking that a couple of things could do with a bit more prominence in the whats-new page.
Specifically:

• minimal sub-package imports are more consistent, but will break user code
  • e.g. import iris; cube = iris.cube.Cube(0)
• speedup for netcdf single-cube loading Simple support for selective netcdf loading. #4176

[pp-mo]

FYI I have now cut the v3.1.0 release : https://github.com/SciTools/iris/releases/tag/v3.1.0

The docs build is up already, and we should shortly have this on conda-forge and PyPI.

Thanks to all involved !

[pp-mo]

3.1.0 is now finalised + up on conda-forge and PyPI
mergeback to 'main' has also been done #4325

[rcomer]

Nice work @pp-mo 🎉