GLPK.Optimizer should be verbose by default, according to MOI docs

[iSoron]

Hi everyone, thank you for your work on this solver interface.

According to MathOptInterface documentation, all optimizers should be verbose by default:

Every optimizer should have verbosity on by default. For instance, if a solver has a solver-specific log level attribute, the MOI implementation should set it to 1 by default. If the user sets Silent to true, then the log level should be set to 0, even if the user specifically sets a value of log level.If the value of Silent is false then the log level set to the solver is the value given by the user for this solver-specific parameter or 1 if none is given.

https://jump.dev/MathOptInterface.jl/stable/reference/models/#MathOptInterface.Silent

GLPK.Optimizer, however, seems to be completely silent by default. For example, the following script produces no output:

using JuMP  # v0.21.8
using GLPK  # v0.14.8

model = Model(GLPK.Optimizer)
@variable(model, x, Bin)
@objective(model, Max, x)
optimize!(model)

Replacing GLPK in the snippet above by Cbc produces some output, as expected.

Should this be considered a bug, and would you consider a PR to fix it?

[odow]

It's currently GLP_MSG_ERR by default. https://github.com/jump-dev/GLPK.jl/blob/5259d51e194e72e9ceee5711d5f56ed1f2077d9d/src/MOI_wrapper/MOI_wrapper.jl#L183 We could bump it up, I guess. Any reason? You can always use set_optimizer_attribute(model, "msg_lev", GLPK.GLP_MSG_ON) (or similar, I forget the exact constant).

[blegat]

Any reason?

It currently does not satisfy the MOI specs that says that it should be on by default.

[odow]

I strongly dislike this change: https://github.com/odow/SDDP.jl/pull/407/checks?check_run_id=2703078270

GLP_MSG_ON is far too verbose. Which is probably why we had GLP_MSG_ERR in the first place.

[blegat]

For algorithms calling a same solvers many times like SDDP, it makes sense to use MOI.Silent

[odow]

Yes, I've turned silent on by default now. The problem is still that GLPK is far too talky. It doesn't print a nice summary, it just spams the console with rows of numbers without context.

I'm not convinced that having the output on by default is an improvement:

julia> model = Model(GLPK.Optimizer)
A JuMP Model
Feasibility problem with:
Variables: 0
Model mode: AUTOMATIC
CachingOptimizer state: EMPTY_OPTIMIZER
Solver name: GLPK

julia> @variable(model, x >= 0)
x

julia> @variable(model, 0 <= y <= 3)
y

julia> @objective(model, Min, 12x + 20y)
12 x + 20 y

julia> @constraint(model, c1, 6x + 8y >= 100)
c1 : 6 x + 8 y ≥ 100.0

julia> @constraint(model, c2, 7x + 12y >= 120)
c2 : 7 x + 12 y ≥ 120.0

julia> print(model)
Min 12 x + 20 y
Subject to
 c1 : 6 x + 8 y ≥ 100.0
 c2 : 7 x + 12 y ≥ 120.0
 x ≥ 0.0
 y ≥ 0.0
 y ≤ 3.0

julia> optimize!(model)
      0: obj =   0.000000000e+00 inf =   2.200e+02 (2)
      3: obj =   2.120000000e+02 inf =   0.000e+00 (0)
*     4: obj =   2.050000000e+02 inf =   0.000e+00 (0)