Remove out-dated precompile.jl and simplify compile_workload

[odow]

There's no point compiling methods for MockOptimzer, and we now use a concrete CachingOptimizer type. If solvers need better recompilation, they can now do it in their package.

As examples, see:

• https://github.com/jump-dev/Ipopt.jl/blob/ffa138dea93d994442b3aa3824688a70c598bacf/src/Ipopt.jl#L40-L99
• https://github.com/jump-dev/HiGHS.jl/blob/80bfb5ad5c735600e2bdc733300f199d74828f52/src/HiGHS.jl#L24-L67

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 1 line in your changes missing coverage. Please review.

Project coverage is 99.00%. Comparing base (b016cdd) to head (0ba2749).
Report is 2 commits behind head on master.

Files with missing lines	Patch %	Lines
src/JuMP.jl	0.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #3871      +/-   ##
==========================================
+ Coverage   98.80%   99.00%   +0.19%     
==========================================
  Files          44       43       -1     
  Lines        6038     6026      -12     
==========================================
  Hits         5966     5966              
+ Misses         72       60      -12     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[odow]

I didn't think this would regress, but it did:

(hgh_precompile) pkg> st
Status `/private/tmp/hgh_precompile/Project.toml`
  [87dc4568] HiGHS v1.11.0
  [4076af6c] JuMP v1.23.3

julia> @time using JuMP, HiGHS
  1.836396 seconds (1.14 M allocations: 80.307 MiB, 2.01% gc time, 0.91% compilation time)

julia> @time begin
         model = Model(HiGHS.Optimizer)
         set_silent(model)
         @variable(model, x >= 0)
         @variable(model, 0 <= y <= 3)
         @objective(model, Min, 12x + 20y)
         @constraint(model, c1, 6x + 8y >= 100)
         @constraint(model, c2, 7x + 12y >= 120)
         optimize!(model)
       end
  0.122040 seconds (77.83 k allocations: 5.237 MiB, 95.64% compilation time)

to

(hgh_precompile) pkg> st
Status `/private/tmp/hgh_precompile/Project.toml`
  [87dc4568] HiGHS v1.11.0
  [4076af6c] JuMP v1.23.3 `https://github.com/jump-dev/JuMP.jl.git#od/precompile`

julia> @time using JuMP, HiGHS
  1.868888 seconds (1.14 M allocations: 79.608 MiB, 1.94% gc time, 0.88% compilation time)

julia> @time begin
         model = Model(HiGHS.Optimizer)
         set_silent(model)
         @variable(model, x >= 0)
         @variable(model, 0 <= y <= 3)
         @objective(model, Min, 12x + 20y)
         @constraint(model, c1, 6x + 8y >= 100)
         @constraint(model, c2, 7x + 12y >= 120)
         optimize!(model)
       end
  0.333175 seconds (201.20 k allocations: 13.715 MiB, 98.47% compilation time)

[odow]

With jump-dev/HiGHS.jl#249 this is back to

julia> @time using JuMP, HiGHS
  1.942927 seconds (1.17 M allocations: 81.800 MiB, 1.32% gc time, 0.92% compilation time)

julia> @time begin
           model = Model(HiGHS.Optimizer)
           set_silent(model)
           @variable(model, x >= 0)
           @variable(model, 0 <= y <= 3)
           @objective(model, Min, 12x + 20y)
           @constraint(model, c1, 6x + 8y >= 100)
           @constraint(model, c2, 7x + 12y >= 120)
           optimize!(model)
       end
  0.093417 seconds (45.31 k allocations: 2.997 MiB, 94.63% compilation time)

[odow]

At the start of the year it was (https://jump.dev/milestones/2024/01/30/jump_year_in_review/)

julia> @time using JuMP, HiGHS
  2.645880 seconds (1.17 M allocations: 82.527 MiB, 4.00% gc time, 0.85% compilation time)

julia> @time begin
         model = Model(HiGHS.Optimizer)
         set_silent(model)
         @variable(model, x >= 0)
         @variable(model, 0 <= y <= 3)
         @objective(model, Min, 12x + 20y)
         @constraint(model, c1, 6x + 8y >= 100)
         @constraint(model, c2, 7x + 12y >= 120)
         optimize!(model)
       end
  0.214786 seconds (241.91 k allocations: 16.071 MiB, 97.47% compilation time: 45% of which was recompilation)

so this is nice progress.

[odow]

For posterity, my other script was:

using SnoopCompileCore
using JuMP, HiGHS
tinf = @snoop_inference begin
    model = Model(HiGHS.Optimizer)
    set_silent(model)
    @variable(model, x >= 0)
    @variable(model, 0 <= y <= 3)
    @objective(model, Min, 12x + 20y)
    @constraint(model, c1, 6x + 8y >= 100)
    @constraint(model, c2, 7x + 12y >= 120)
    optimize!(model)
end;
using SnoopCompile, ProfileView
fg = flamegraph(tinf)
ProfileView.view(fg)