Oscar Dowson

`initialize` is essentially like `final_touch`. It says "I've finished building the model, and I'm ready to set everything up." But the problem is that we want people to be able...

So one way to move forward with this is to update MOI.Nonlinear so that `initialize` doesn't store expression graphs. Then we can special-case JuMP to avoid setting the evaluator if...

Part of the problem is that we have an `Evaluator` which requires a fixed model. (You can't add new variables after creating the `Evaluator` object.) We could switch to some...

This allows a loop like: ```Julia for i in 1:n for (i,load) in data["load"] set_value(pd_parameter[load["index"]], (1.0+(rand()-0.5)/10)*load["pd_base"]) set_value(qd_parameter[load["index"]], (1.0+(rand()-0.5)/10)*load["qd_base"]) end if i > 1 x = all_variables(model) x0 = value.(x) set_start_value.(x,...

Currently blocked by https://github.com/domluna/JuliaFormatter.jl/issues/603 and https://github.com/MichaelHatherly/CommonMark.jl/issues/43.

Current JuMP failure is ```Julia (jump-tests-m1) pkg> test JuMP Testing JuMP Status `/private/var/folders/tz/_3rffzvd1hs3rrxssmntjnpw0000gn/T/jl_Pxer0Z/Project.toml` [49dc2e85] Calculus v0.5.1 [864edb3b] DataStructures v0.18.11 [f6369f11] ForwardDiff v0.10.27 [682c06a0] JSON v0.21.3 [4076af6c] JuMP v1.0.0 [b8f27783] MathOptInterface...

Native M1. Rosetta works fine.

Some progress. JuMP and HiGHS are now passing tests on M1 with Julia v1.8-rc4. The COIN-OR suite of Cbc, Clp, and Ipopt are still segfaulting and/or failing.

Ipopt is failing one test (norm limit, instead of finding an optimal solution) that I haven't managed to get to the bottom of, but it is passing all others, so...

Closing because the JuMP side is working fine in Julia v1.8. Cbc and Clp can track progress in their own issues.