Kristoffer Carlsson

I don't really think this is the right way to go. It wouldn't handle cases like ``` ... big_loop() breakpoint ``` People would need to start splitting up their functions...

> This is not surprising as some of the print/show commands end up running infinite-loops for these particular types (infinite arrays). It kinda feels like the type should make `print`...

> It works fine on the REPL We call `repr` on it. > (Also, the point of a debugger is to work with buggy code....) And it made you find...

Probably around https://github.com/JuliaDebug/Debugger.jl/blob/7e3ebd16fbc8524f03d88cd7b64ccbcdfa7add7d/src/repl.jl#L232-L238

I took a stab at `teh/compiled_ccall` and where I am kinda stuck at now is for functions that look like ```jl unsafe_convert(::Type{Ptr{T}}, a::Array{T}) where {T} = ccall(:jl_array_ptr, Ptr{T}, (Any,), a)...

What is your opinion about pushing things to `compiled_methods` that are very commonly used but unlikely to be of interest for breakpoints. For example, `iterate(a::Array)` (with `Core_apply` now being stepped...

Here is a fun one: ```jl julia> using JuliaInterpreter, Plots julia> @time @interpret plot(rand(5)) 5.703063 seconds (39.36 M allocations: 2.545 GiB, 6.56% gc time) ``` vs ```jl julia> using JuliaInterpreter,...

I haven't profiled yet, just observed it.

I think this is the problem: ```jl julia> using BenchmarkTools julia> c = Some{Any}("foo") Some("foo") julia> @code_warntype convert(Some{Any}, c) Body::Some{Any} 1 ─ %1 = (Base.getfield)(x, :value)::Any │ %2 = %new(Some{Any},...

With https://github.com/JuliaLang/julia/pull/31536 we are at ``` julia> @time @interpret plot(rand(5)) 2.930957 seconds (13.54 M allocations: 481.296 MiB, 2.31% gc time) ``` for the non CSV case and ``` julia> @time...