Mirror of forum.nim-lang.org

1228 :: Arithmetic on static[int] (compiler crash)

• andrea (orginal) [2015-05-14T15:00:50+02:00] view original

  I have tried defining types parametric in a static[int], and do arithmetic on this numbers. Turns out, this does not work.

  What is an issue, though, is that the compiler does not report an error, nor does it manage to compile, but silently crashes. This happen when compiling, for instance

  type
    Foo[a: static[int]] = array[a, int]
    Bar[a, b: static[int]] = array[b, Foo[a/2]]
  
  echo "done"
  Run

  When running nim c example.nim I do not get any error, but no executable is produced as well.

---

• skyfex (orginal) [2015-07-28T22:09:55+02:00] view original

  I found this question and wondered if it had ever been resolved.

  For future reference, it seems it has been registered as an issue at Github:

  https://github.com/nim-lang/Nim/issues/2730

  (Just to help people searching the forums)

---

• gmfawcett (orginal) [2015-07-29T01:37:49+02:00] view original

  For what it's worth, I tried compiling the code with a debug build of the compiler; I got a stacktrace, but no error message. Changing line 1600 of vmgen.nim from

  internalError(n.info, "cannot generate code for: " & s.name.s)
  Run

  to

  globalError(n.info, "cannot generate code for: " & s.name.s)
  Run

  and then recompiling the compiler, I could then compile the test code at --verbosity:2 and got the message:

  c.nim(5, 40) Error: cannot instantiate Foo
  got: (float)
  but expected: (a)
      Bar[a, b: static[int]] = array[b, Foo[a/2]]
  Run

  I don't know nearly enough about the internals to suggest a fix, but the compiler "crash" seems to be a problem with the error reporting, and not with the codegen. [edit: this doesn't address the typing issue, of course, just the apparent compiler crash.]

---

• andrea (orginal) [2015-07-29T14:09:54+02:00] view original

  You are right that / is the wrong operation. Still, even when using, say +, the code does not compile.

  As of today, though, the error reporting seems correct.

  When I reported the issue, changing the inline operation to a proc would make everything work. As of today, this seems not to work anymore

---