Oscar Dowson

Convex.jl lacks support for a lot of broadcasting operations. This causes a bunch of problems like this: ```Julia julia> using Convex julia> x = Variable(2) Variable size: (2, 1) sign:...

Juniper reports the `PrimalStatus` as `INFEASIBLE_POINT` when the problem is not optimal: https://github.com/lanl-ansi/Juniper.jl/blob/053661ebc023ff4534210207f9ca40c0ff42fca0/src/MOI_wrapper/results.jl#L29-L39 This is incorrect, because sometimes the point is actually feasible. If we can prove feasibility, it should...

See https://discourse.julialang.org/t/how-to-write-conditional-expression-using-jump/83106 The trace is ```Julia MethodError: no method matching get_branching_disc_idx!(::Juniper.JuniperProblem, ::Juniper.StepObj, ::Juniper.SolverOptions, ::Vector{Int64}, ::Juniper.GainObj, ::Int32) Closest candidates are: get_branching_disc_idx!(::Any, ::Any, ::Any, ::Any, ::Any) at ~/.julia/packages/Juniper/0Z1vO/src/BnBTree.jl:14 get_branching_disc_idx!(::Any, ::Any, ::Any, ::Any,...

There are a few MOI tests that fail: https://github.com/lanl-ansi/Juniper.jl/blob/ec81a1014bfe3c8e3b648fe7a016494626de0ea8/test/MOI_wrapper.jl#L57-L72 This could be bugs in Juniper, or bugs in the MOI wrapper.

format_docstrings is not idempotent

2

Another example from https://github.com/jump-dev/JuMP.jl/pull/2993 ```julia s = """ \"\"\" solve_knapsack( optimizer, data_filename::String, config::_AbstractConfiguration, ) Solve the knapsack problem and return the optimal primal solution ## Arguments - `optimizer`: an object...

Inconsistent white-space around binary operators

8

[JuMP's style guide](https://www.juliaopt.org/JuMP.jl/v0.20.0/style/#Whitespace-1) suggests that all binary operators (with two exceptions) are surrounded by spaces. The two exceptions are `^` and `:`. So `x^2` not `x ^ 2` and `1:n`...

Strip return nothing

6

JuMP's style guide prefers ```Julia function foo() return end ``` over ```julia function foo() return nothing end ``` Currently, the formatter just leaves whatever was there. Is it possible to...

Objective state errors are bad

1

If you specify `lower_bound` but not `upper_bound`, then bad things happen.

Some feedback from a user in my emails (search "Suggestions for JuMP tutorial" 28 February). https://odow.github.io/SDDP.jl/stable/tutorial/basic/07_arma/ https://odow.github.io/SDDP.jl/stable/tutorial/advanced/11_objective_states/ ``` Doubts from the Vector auto-regressive model, model = SDDP.LinearPolicyGraph( stages = 3,...