Dict display performance

[jakebolewski]

I have not looked into why this is the case but here is an example:

julia> D = Dict()
Dict{Any,Any} with 0 entries

julia> for key in [randstring() for i=1:5000]
           D[key] = randn(100_000)
       end

julia> @time display(D)
  "kfCodtXI" => [1.6888401174417957,-1.4307288883093494,0.9559295172958958,0.7167676654234264,0.3473601699748082,0.9725759548997545,-0.4779459387535729,-0.9965805074751564,-0.552766042152107,1.5609540447598185,0.409914342350185,0.20369490772349627,…
  ⋮          => ⋮
elapsed time: 1.716493913 seconds (441943508 bytes allocated, 1.54% gc time)

vs

julia> A = randn((5_000, 100_000));
julia> @time display(A);
5000x100000 Array{Float64,2}:
elapsed time: 0.01819924 seconds (3621724 bytes allocated)

[JeffBezanson]

It might be because it uses sprint(show, v) on every value, which prints all of it to a string, and then truncates afterwards.

[jakebolewski]

Makes sense. This is really bad for large sparse arrays, see #10257.

[mbauman]

Dup of #9481, and really depends upon #7959 for a proper fix.

[JeffBezanson]

Fixing performance this terrible doesn't depend on any big redesigns.