2265 rename split dyn stencil

[oakleybrunt]

closes #2265

sub-issue of #2235

[codecov]

Codecov Report

Patch coverage: 100.00% and no project coverage change.

Comparison is base (5a77078) 99.85% compared to head (9fba2af) 99.85%.
Report is 1 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #2266   +/-   ##
=======================================
  Coverage   99.85%   99.85%           
=======================================
  Files         339      340    +1     
  Lines       46027    46030    +3     
=======================================
+ Hits        45958    45961    +3     
  Misses         69       69           

Files Changed	Coverage Δ
src/psyclone/domain/lfric/__init__.py	100.00% <100.00%> (ø)
src/psyclone/domain/lfric/lfric_stencil.py	100.00% <100.00%> (ø)
src/psyclone/dynamo0p3.py	99.82% <100.00%> (-0.01%)	⬇️

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

[arporter]

Thanks for doing this @oakleybrunt. I think it raises the question about whether we even need a separate class for this so I will gather opinions and then we can decide what to do.

[arporter]

Looking at this class, it seems to me we could do without it and have a NamedTuple instead:

LFRicArgStencil = namedtuple("Symbol", "name extent extent_arg direction_arg")

I say this because none of the setters do any validation. Can @rupertford, @TeranIvy or @hiker see any potential problem with doing that? (e.g. is this class likely to need significant additional functionality in future?)

[TeranIvy]

@arporter, I think that the potential future functionality would be "more of the same" in the sense of more stencil types, but the underlying mechanism wouldn't change. If doing away with the class doesn't affect adding more stencil types then I'm fine with it!

[arporter]

I've assigned everyone who I think might care so I can get your attention :-) The question is, is everyone happy if we do away with having a separate class (and tests etc) for 'DynStencil' in favour of just having a namedtuple?

[arporter]

Hi @oakleybrunt, don't do anything yet - in all likelihood we will delete the class entirely!

[oakleybrunt]

Hi @oakleybrunt, don't do anything yet - in all likelihood we will delete the class entirely!

I had worked my way down the list before reading all the comments!! When I got to your final comment about changing to a namedtuple I had already done the edits 😆 I thought I may as well commit and push them. Thanks!

[arporter]

OK, no one has complained so I suggest we go with the namedtuple route. If you look in psyclone/domain/lfric/lfric_types.py you'll see a few examples. I suggest you cut a new branch for this in case it goes pear shaped and we decide that what you've done in this PR is actually better. (The new namedtuple can stay in the dynamo0p3.py file.)

[oakleybrunt]

@arporter What are we doing with this PR? Is this closed now?

[arporter]

This proposed change was replaced with the use of a dataclass in #2321.