compiler/nimeval can't be used twice: fails 2nd time with: Error: internal error: n is not nil

[timotheecour]

/cc @Araq
env var nimc_D points to compile git repo

import compiler / [ast, vmdef, vm, nimeval, llstream]
from os import getEnv
proc evalString(code: string, moduleName = "script.nim"):auto=
  let stream = llStreamOpen(code)
  defer: llStreamClose(stream)
  let path = getEnv("nimc_D") & "/lib/"
  var intr = createInterpreter(moduleName, [path])
  defer: destroyInterpreter(intr)
  intr.evalScript(stream)

evalString("echo 1")
evalString("echo 2")

rnim bugs/nimeval/t03_n_is_not_nil.nim
nim c --nimcache:/tmp/nim//nimcache/ -o:/tmp/nim//app -r bugs/nimeval/t03_n_is_not_nil.nim
1
???(0, 0) Error: internal error: n is not nil
Traceback (most recent call last)
/Users/timothee/git_clone/nim/timn/bugs/nimeval/t03_n_is_not_nil.nim(13) t03_n_is_not_nil
/Users/timothee/git_clone/nim/timn/bugs/nimeval/t03_n_is_not_nil.nim(8) evalString
/Users/timothee/.nimble/pkgs/compiler-0.19.0/compiler/nimeval.nim(115) createInterpreter
/Users/timothee/.nimble/pkgs/compiler-0.19.0/compiler/modules.nim(107) compileSystemModule
/Users/timothee/.nimble/pkgs/compiler-0.19.0/compiler/modules.nim(71) compileModule
/Users/timothee/.nimble/pkgs/compiler-0.19.0/compiler/passes.nim(220) processModule
/Users/timothee/.nimble/pkgs/compiler-0.19.0/compiler/passes.nim(95) closePasses
/Users/timothee/.nimble/pkgs/compiler-0.19.0/compiler/sem.nim(626) myClose
/Users/timothee/.nimble/pkgs/compiler-0.19.0/compiler/msgs.nim(524) internalError
/Users/timothee/.nimble/pkgs/compiler-0.19.0/compiler/msgs.nim(495) liMessage
/Users/timothee/.nimble/pkgs/compiler-0.19.0/compiler/msgs.nim(324) handleError
/Users/timothee/.nimble/pkgs/compiler-0.19.0/compiler/msgs.nim(314) quit
Error: execution of an external program failed: '/tmp/nim//app '

EDIT
/cc @alaviss
looks suspiciously similar to [1] which was fixed by @alaviss

[1] running runNimScript on 2 subsequent nims files causes "Error: internal error: n is not nil" #8235

[alaviss]

See #9068

[timotheecour]

yes; I saw the thread, a better fix may indeed be to address #9068; but I'll let someone more familiar with that code do the proper fix; until then #9181 fixes this issue so I'd like it to be merged

[alaviss]

@timotheecour Can you confirm that #9259 fixed this issue? If so please close it.

[timotheecour]

it fixes it but there's still not test case for it; I've just renamed #9181 to reflect the fact that #9259 was merged in the meantime; I think it's better to close once the test case is merged, so let's wait for #9181 to close this