Different outputs between CUTEst.select & select_sif_problems

[alonsoC1s]

Trying to reproduce the problem selection section in the tutorial for a PR I noticed the output from the deprecated CUTEst.select and the new select_sif_problems differs significantly in length (and thus content).

With the old select:

julia> old_p = CUTEst.select(contype = "unc")
293-element Vector{String}:
 "LIARWHD"
 "SCHMVETT"
 "LUKSAN13LS"
 ⋮

With select_sif_problems:

julia> new_p = select_sif_problems(max_con=0)
476-element Vector{String}:
 "TRIGON2B"
 "LIARWHD"
 "SCHMVETT"
 "LUKSAN13LS"
 ⋮

Initially I thought I might be misusing the new selection tool but found examples of problems that are wrongly selected.
I believe there might both be, problems that do have constraints and are incorrectly selected by the deprecated version

julia> setdiff(old_p, new_p)
1-element Vector{String}:
 "S308NE"
julia> p = CUTEstModel("S308NE")
  Problem name: S308NE
   All variables: ████████████████████ 2      All constraints: ████████████████████ 3     
            free: ████████████████████ 2                 free: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0     
           lower: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0                lower: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0     
           upper: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0                upper: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0     
         low/upp: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0              low/upp: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0     
           fixed: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0                fixed: ████████████████████ 3     
          infeas: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0               infeas: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0     
            nnzh: (  0.00% sparsity)   3               linear: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0     
                                                    nonlinear: ████████████████████ 3     
                                                         nnzj: ( 33.33% sparsity)   4   

as well as unconstrained problems that should be selected and are not

julia> setdiff(new_p, old_p)
184-element Vector{String}:
 "TRIGON2B"
"DECONVB"
 "POWELLBC"
 ⋮
julia> p = CUTEstModel("DECONVB")
  Problem name: DECONVB
   All variables: ████████████████████ 63     All constraints: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0     
            free: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0                 free: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0     
           lower: █████████████⋅⋅⋅⋅⋅⋅⋅ 40               lower: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0     
           upper: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0                upper: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0     
         low/upp: ████⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 11             low/upp: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0     
           fixed: ████⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 12               fixed: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0     
          infeas: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0               infeas: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0     
            nnzh: ( 44.89% sparsity)   1111            linear: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0     
                                                    nonlinear: ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅ 0     
                                                         nnzj: (------% sparsity)         

[amontoison]

@alonsoC1s
select and select_sif_problems is exactly the same function:
https://github.com/JuliaSmoothOptimizers/CUTEst.jl/blob/main/src/classification.jl#L161

The issue is more related to the different behavior with options max_con=0 and "contype = unc".

I think that one option allows just bound-constrained problem and the other one pure unconstrained problems.

[alonsoC1s]

I might be a bit confused with the problem summary then. Taking S308NE as an example, the summary says that there are 3 constraints. Should that problem then be selected by either max_con=0 or "contype = unc"?