Add find_type_instabilities

[davidanthoff]

Usage is:

Mimi.find_type_instabilities(m, :compname)

and that will call code_warntype on the run_timestep method for that component.

In theory one can also use this with https://github.com/JuliaDebug/Cthulhu.jl by doing

Mimi.find_type_instabilities(m, :compname, Cthulhu.descend)

but in practice that doesn't seem to work together with our way of generating these run_timestep methods from our macro.

I would suggest we don't document this for now and don't consider it part of the "public" API, until we have figured out whether this is actually working well or not. We might also try to get this to work with Cthulhu properly.

[codecov]

Codecov Report

Patch coverage has no change and project coverage change: -0.28 ⚠️

Comparison is base (91a71f1) 85.12% compared to head (4bdae7a) 84.85%.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #961      +/-   ##
==========================================
- Coverage   85.12%   84.85%   -0.28%     
==========================================
  Files          40       41       +1     
  Lines        4041     4054      +13     
==========================================
  Hits         3440     3440              
- Misses        601      614      +13     

Flag	Coverage Δ
unittests	84.85% <0.00%> (-0.28%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
src/Mimi.jl	100.00% <ø> (ø)
src/utils/diagnostics.jl	0.00% <0.00%> (ø)

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[lrennels]

I was getting an error because the values of the dim_dict are nothing for the components, so altered the access to data structures ... I think this will work better but need to double check time

[davidanthoff]

I only tried this after running the model, maybe at that point the dims are there? But if we can get this to work without running, even better.

[lrennels]

Interesting, at first I was doing

m = get_model
Mimi.build!(m)

to get a built model but not run it because the migration model is just so slow, but I also tried it with a run MimiGIVE and got the same error? Did yours work for a run model?

[lrennels]

I think we want to put a AbstractTimestep type ie. Mimi.FixedTimestep{1750, 1, 2300} in t to get better results ... I experimented it with it and I think it works well.

We should be able to get that from the clock or something, that is parameterized by the Timestep type I think, I'll take a look tomorrow.

[lrennels]

@davidanthoff looks good, I've tested it quite a bit in practice and it's extremely helpful! Agree on keeping it out of docs and mostly with developers until we're certain the kinks are worked out.