Thaaat sounds like a breaking change to me 😣

Need to change

Oceananigans = "0.106.3"

Thaaat sounds like a breaking change to me 😣

oof sorry. We used to interpret "breaking change" as "breaking to the user interface / exported functions". we need to change the Oceananigans convention to reflect its status as a library that other packages depend on, now, who may depend on specific aspects of the internal implementation.

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

Funnily enough, I was talking about this with @jtbuch earlier today 😅 Semver is nice because it gives some guidelines, but then it leaves to interpretation what constitutes "public API", or even worse what's a "breaking change" vs a "bugfix", and people can see things differently.

You're correct that a legitimate interpretation of "public API" could be "exported functions", some people include "documented (not necessarily exported) functions". Julia v1.11 has a public keyword to unambiguously and programmatically identify public API (although that's usually considered sufficient to identify public API, but not necessary, because it isn't widely used yet). But this case the affected function was one which was created almost only for consumption in downstream packages like Breeze 😅

Also we probably should not define new time-steppers here, but rather have general implementations in Oceananigans for all models to use. The SSPRK3 is currently in Breeze only to accelerate development as we may need to modify it for atmosphere-specific things (which are unknown yet). But in the long run it should be moved to Oceananigans I think, because it can also be used by Oceananigans.NonhydrostaticModel (for example)...

But also there is an interface for defining time-steppers, so perhaps in general we should regard that as public.