Oscar Dowson
Oscar Dowson
Perhaps I'll let @blegat speak to whether it is a good fit because this package is really his. But note that MA is not really a generic interface for all...
Is this an improvement? I'm not sure. Now: https://jump.dev/JuMP.jl/previews/PR3821/tutorials/linear/multi/ Previous: https://jump.dev/JuMP.jl/stable/tutorials/linear/multi/
So if there are multiple caches and bridges, you want the last cache before the optimizer? Even if we added this, it's complicated to communicate to a user. This warning...
I still think this is the wrong approach. There's a lot of complexity here, and having more than one concept for what the unsafe backend is is going to cause...
How about something like this: ```Julia julia> using JuMP, SCS julia> function optimizer_layers(model::GenericModel) return _optimizer_layers!(backend(model)) end optimizer_layers (generic function with 1 method) julia> _backend(::MOI.ModelLike) = nothing _backend (generic function with...
I'm still not convinced that we need this.
I thought the conclusion was that we should add the `optimizer_layers` function?
I think my conclusion with this is that we don't need it.
I think we decided not to do this. We have some better support with `set_macro_timing `, but this tutorial could still be improved: https://jump.dev/JuMP.jl/dev/tutorials/getting_started/debugging/#Debugging-performance-problems
I don't know what the right approach is, but after thinking about this more I don't think that it's this PR.