Debugger cannot handle calls to R made using RCall

[PGS62]

Debugger fails when attempting to debug calls to RCall.reval as described in this post I made today on Julia Discourse. It was suggested that I raise it as an issue here.

Here is the example posted:

using RCall
function test()
    reval("1+1")
end

julia> test()
RObject{RealSxp}
[1] 2


julia> @enter test()
In test() at d:\Philip\Source\script\Julia\JSource\xccyHWhooks.jl:373
 372  function test()
>373      reval("1+1")
 374  end

About to run: (RCall.reval)("1+1")
1|debug> n
ERROR: MethodError: no method matching iterate(::Type{Ptr{RCall.SxpHead}})
Closest candidates are:
  iterate(::Core.SimpleVector) at essentials.jl:568
  iterate(::Core.SimpleVector, ::Any) at essentials.jl:568
  iterate(::ExponentialBackOff) at error.jl:199
  ...
Stacktrace:
 [1] safe_parseVector(::Ptr{StrSxp}, ::Base.RefValue{Int32}, ::Ptr{NilSxp}) at C:\Users\Philip\.julia\packages\RCall\g7dhB\src\eval.jl:6
 [2] rparse_p(::Ptr{StrSxp}, ::Ptr{NilSxp}) at C:\Users\Philip\.julia\packages\RCall\g7dhB\src\eval.jl:49
 [3] rparse_p(::String, ::Ptr{NilSxp}) at C:\Users\Philip\.julia\packages\RCall\g7dhB\src\eval.jl:70
 [4] rparse_p(::String) at C:\Users\Philip\.julia\packages\RCall\g7dhB\src\eval.jl:70
 [5] reval(::String, ::RObject{EnvSxp}) at C:\Users\Philip\.julia\packages\RCall\g7dhB\src\eval.jl:132
 [6] reval(::String) at C:\Users\Philip\.julia\packages\RCall\g7dhB\src\eval.jl:132
 [7] test() at d:\Philip\Source\script\Julia\JSource\xccyHWhooks.jl:373

[timholy]

This is probably a dup of #285. I can't figure out how to resolve the module-hygeine issues in the potential fix, #290. Maybe we need to disable building compiled ccall etc for parametric functions?

[KristofferC]

Maybe we need to disable building compiled ccall etc for parametric functions?

Is there any way we can detect the bad case and disable compilation only in those cases?

[timholy]

Maybe RetType.hasfreetypevars && return nothing? Haven't tested it.

[timholy]

Might be fixed by #290. I can't install RCall properly so I can't easily check.

[timholy]

#290 is merged now so please try the master branch and close this if it is fixed.

[PGS62]

I think the problem is not fixed. To reach that conclusion I did the following:

1. Verified that I could replicate the problem I reported at the start of this thread.
2. Ran update at the package prompt.
3. Also ran add JuliaInterpreter. Previously JuliaInterpreter was not listed by a call to status even though Debugger was listed and I understand that Debugger uses JuliaInterpreter.
4. Noted that status now listed JuliaInterpreter as version v0.7.4 which is the version released October 3.
5. Re-ran the test, but saw the same failure - see below.

using RCall
function test()
    reval("1+1")
end

julia> test()
RObject{RealSxp}
[1] 2


julia> @enter test()
In test() at d:\Philip\Source\script\Julia\JSource\adhoc2.jl:22
 21  function test()
>22      reval("1+1")
 23  end

About to run: (RCall.reval)("1+1")
1|debug> n
ERROR: MethodError: no method matching iterate(::Type{Ptr{RCall.SxpHead}})
Closest candidates are:
  iterate(::Core.SimpleVector) at essentials.jl:568
  iterate(::Core.SimpleVector, ::Any) at essentials.jl:568
  iterate(::ExponentialBackOff) at error.jl:199
  ...
Stacktrace:
 [1] safe_parseVector(::Ptr{StrSxp}, ::Base.RefValue{Int32}, ::Ptr{NilSxp}) at C:\Users\Philip\.julia\packages\RCall\g7dhB\src\eval.jl:6
 [2] rparse_p(::Ptr{StrSxp}, ::Ptr{NilSxp}) at C:\Users\Philip\.julia\packages\RCall\g7dhB\src\eval.jl:49
 [3] rparse_p(::String, ::Ptr{NilSxp}) at C:\Users\Philip\.julia\packages\RCall\g7dhB\src\eval.jl:70
 [4] rparse_p(::String) at C:\Users\Philip\.julia\packages\RCall\g7dhB\src\eval.jl:70
 [5] reval(::String, ::RObject{EnvSxp}) at C:\Users\Philip\.julia\packages\RCall\g7dhB\src\eval.jl:132
 [6] reval(::String) at C:\Users\Philip\.julia\packages\RCall\g7dhB\src\eval.jl:132
 [7] test() at d:\Philip\Source\script\Julia\JSource\adhoc2.jl:22


julia> versioninfo()
Julia Version 1.1.1
Commit 55e36cc308 (2019-05-16 04:10 UTC)
Platform Info:
  OS: Windows (x86_64-w64-mingw32)
  CPU: Intel(R) Core(TM) i7-6700 CPU @ 3.40GHz
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-6.0.1 (ORCJIT, skylake)
Environment:
  JULIA_EDITOR = "C:\Users\Philip\AppData\Local\Programs\Microsoft VS Code\Code.exe"

(v1.1) pkg> status
    Status `C:\Users\Philip\.julia\environments\v1.1\Project.toml`
  [6e4b80f9] BenchmarkTools v0.4.3
  [9a962f9c] DataAPI v1.1.0
  [82cc6244] DataInterpolations v1.1.1
  [31a5f54b] Debugger v0.6.2
  [39dd38d3] Dierckx v0.4.1
  [31c24e10] Distributions v0.21.3
  [a98d9a8b] Interpolations v0.12.2
  [682c06a0] JSON v0.21.0
  [aa1ae85d] JuliaInterpreter v0.7.4
  [2774e3e8] NLsolve v4.1.0
  [86f7a689] NamedArrays v0.9.3
  [91a5bcdd] Plots v0.26.3
  [6f49c342] RCall v0.13.4
  [295af30f] Revise v2.2.0
  [f2b01f46] Roots v0.8.3
  [ed01d8cd] Sobol v1.3.0
  [bd369af6] Tables v0.2.11
  [9ff05d80] TickTock v1.0.0
  [44d3d7a6] Weave v0.9.1
  [8bb1440f] DelimitedFiles

[anna-parker]

Just wanted to second that I also believe that this problem is not fixed in VScode. Strangely enough RCall works for me in the VS code debug console but not when called in a script in debugger mode. For me the trackback also always leads back to the rparse_p function, as in your example.

When just testing using the simple example
import Pkg; Pkg.add("RCall") Pkg.build("RCall") using RCall R"rnorm(10)"

I get the trace back:

TypeError: in typeassert, expected Core.SimpleVector, got a value of type Int64

Stacktrace:
 [1] protect(p::Ptr{StrSxp})
   @ RCall [~/.julia/packages/RCall/6kphM/src/types.jl:269]()
 [2] rparse_p(st::Ptr{StrSxp}, sf::Ptr{NilSxp})
   @ RCall [~/.julia/packages/RCall/6kphM/src/eval.jl:49]()
 [3] rparse_p(st::String, sf::Ptr{NilSxp})
   @ RCall [~/.julia/packages/RCall/6kphM/src/eval.jl:74]()
 [4] rparse_p(st::String)
   @ RCall [~/.julia/packages/RCall/6kphM/src/eval.jl:74]()
 [5] reval(str::String, env::RObject{EnvSxp})
   @ RCall [~/.julia/packages/RCall/6kphM/src/eval.jl:136]()
 [6] top-level scope
   @ [~/Documents/ARGTools/plot_tester.jl:5]()

However, this works via the debug console, jupyter notebooks and when run via the command line.