Closed AayushSabharwal closed 5 days ago
Eventually I'll move these to SII's downstream testset
this looks great! one question is now that all these getindexes just call getu, could they all be merged into a single method?
I'm happy to give it a shot, but defining these methods with a Union
doesn't feel right to me. Semantically the fact that problems, integrators and solutions all support getindex
is completely separate (they're four different interfaces) and basically just a coincidence.
Also, setindex!
for problems uses ___internal_setindex!
for reasons I'm not aware of, so we'd just end up merging time-independent solutions and DEIntegrator
.
The tests have been quite heavily modified, so I cannot really tell if they are the same or no. If you say they are the same I trust you.
How is this going?
The added tests pass locally last I checked, but the core CI is failing due to an error I haven't been able to reproduce yet (even with --check-bounds=yes
and --depwarn=yes
). That code hasn't been touched, so the sudden failure is weird.
In fact, even in downstream CI all the depwarns are erroring. I guess tests are now running with --depwarn=yes
for some reason?
@TorkelE I moved your symbolic indexing tests here, added some more, and replaced
symbol_indexing.jl
. I've ensured everything insymbol_indexing.jl
is still tested somewhere. Tests pass using MTK masterChecklist
Additional context
Add any other context about the problem here.