test(Rosenbrock): update resize_tests for new cache fields (k₁/k₂/k₃ → ks, f₁ → dense)#3313
test(Rosenbrock): update resize_tests for new cache fields (k₁/k₂/k₃ → ks, f₁ → dense)#3313singhharsh1708 wants to merge 2 commits intoSciML:masterfrom
Conversation
… → ks, f₁ → dense)
| @test length(i.cache.k₁) == 5 | ||
| @test length(i.cache.k₂) == 5 | ||
| @test length(i.cache.k₃) == 5 | ||
| @test all(length.(i.cache.ks) .== 5) |
There was a problem hiding this comment.
also test that there is the right number of ks.
That's only with your PR though? So this fails without your PR. Is there a reason this isn't in the PR branch? |
I’ve kept the Rosenbrock23/32 → RodasTableau migration PR focused on code changes only and removed all test updates from it as suggested.
The test changes are now in a separate PR since they depend on the new RosenbrockCache structure (k₁/k₂/k₃ → ks, f₁ → dense). This PR is intended to be merged after the migration PR.
|
No, unrelated test changes should be a separate PR. This is a test that must change because of your changes. Your changes change it from using k1, k2, k3 to a ks vector, so the update should be together. This cannot merge without that change because it would fail as it's simply not correct to change to a ks vector without moving the source code. |
2 participants
Summary
Updates
resize_tests.jlto reflect the newRosenbrockCachestructure introduced in the RodasTableau refactor.The previous tests accessed fields specific to
Rosenbrock23Cache/Rosenbrock32Cache:k₁,k₂,k₃f₁These have been replaced in the unified
RosenbrockCachewith:ks(vector of stage values)dense(interpolation coefficients)Changes
Replaced:
k₁,k₂,k₃→ksf₁→denseUpdated assertions to validate vectorized structure:
all(length.(cache.ks) .== N)all(length.(cache.dense) .== N)Rationale
The cache structure was unified as part of the RodasTableau migration, so tests need to align with the new representation.
No changes to solver behavior — this only updates tests to match the new internal structure.
Test for #3273
Checklist