Removal of dependency on SciMLBase.ODESolution

[mateuszbaran]

This will be the proper resolution of issues caused by abusing internals of OrdinaryDiffEq.jl. We will still use parts of SciML that make sense to keep using but there are parts we can't use such as SciMLBase.ODESolution.

TODO:

• Adapt parts of solve that are necessary.
• Check coverage.
• Write changelog entry.

[kellertuer]

Nice! Looking forward to that. For me it is not yet fully clear which parts come from where, and whether we would still have a nice solve!-tingy, but I am confident you already see a way that our examples would not have to be changed too much.

[mateuszbaran]

Examples and API shouldn't have to change much or even at all. This is mostly an internal rework.

[codecov]

Codecov Report

Attention: Patch coverage is 86.04651% with 24 lines in your changes missing coverage. Please review.

Project coverage is 92.59%. Comparing base (af86f91) to head (7cadf45).
Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
src/ManifoldDiffEq.jl	86.33%	22 Missing ⚠️
src/interpolation.jl	50.00%	1 Missing ⚠️
src/problems.jl	50.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #33      +/-   ##
==========================================
- Coverage   95.12%   92.59%   -2.53%     
==========================================
  Files           8        8              
  Lines         328      486     +158     
==========================================
+ Hits          312      450     +138     
- Misses         16       36      +20     

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

[kellertuer]

Does this make #28 obsolete by the way?

[mateuszbaran]

Yes, completely.

[kellertuer]

Sorry for quite a few comments. This looks like a very good step in the right direction!

[kellertuer]

Ah and we could update the readme, since we no longer directly depend on OrdinaryDiffEq after this PR.

[mateuszbaran]

Ah and we could update the readme, since we no longer directly depend on OrdinaryDiffEq after this PR.

We still depend on OrdinaryDiffEqCore.jl for timestepping and OrdinaryDiffEqCore.jl lives in the OrdinaryDiffEq repository so the readme isn't technically wrong here? Anyway, feel free to rephrase it if you like.

[kellertuer]

I did not check that detailed, then it can stay as is for now I think. I feel the dependency is now a bit more lightweight, but that maybe has to be very carefully phrased, since we still aim to fit into the framework.

[mateuszbaran]

The dependency is indeed more lightweight. After this PR it's mostly timestepping management and the fact that we follow the same interface.

[kellertuer]

Well, I trust you that you will update the changelog and care for a bit of code coverage :)

[mateuszbaran]

I've created the changelog, I'm not sure why the check is failing -- probably because there is no changelog on the main branch?

Code coverage is more or less as good as it can get without testing things we weren't testing before so I don't really care that much.

[mateuszbaran]

Ah, news file was in the wrong folder.