Mirror of forum.nim-lang.org

3418 :: [solved] how do I set current directory with os.setCurrentDir(newDir: string) ?

• refaqtor (orginal) [2017-12-13T04:05:17+01:00] view original

  [solution] it turns out that I shouldn't use 'discard' where nothing is returned

  ---

  I'm getting an error that I don't understand when using setCurrentDir.

  I try setting the current directory

  discard setCurrentDir(newDir: "/path/that/already/exists")
  Run

  which results in

  Error: type expected

  As a 2 day user of the language, I'm not sure what to make of that message.

  Any hints? Thanks!

  ---

  I managed to execute processes in particular working directories by calling osproc.startProcess() directly

  proc runProcess*(cmd: string; wrkdir: string = ""): string =
      var arg: set[ProcessOption]
      arg = { poEvalCommand }
      let runningProcess = startProcess(command = cmd, workingDir = wrkdir, options = arg)
  Run

  which, ultimately, gives me more control. But, misses the convenience of nim that I had enjoyed until now. Anyway, onward!

---

• dom96 (orginal) [2017-12-14T12:40:29+01:00] view original

  I'd love to make this error message better but it's difficult to see how to make it better for someone that's been using Nim for so long.

  Let me try to explain this error:

  let x: int = 4
  Run

  The int is a type. In Nim : is usually followed by a type, hence "Error: type expected". Of course, you can also construct types like this:

  type
    Obj = object
      x: int
  
  echo Obj(x: 42)
  Run

  But you cannot call procedures like this. You use = for that instead.

  discard setCurrentDir(newDir="/path/that/already/exists")
  Run

  or just

  discard setCurrentDir("/path/that/already/exists")
  Run

  If you've got suggestions on how to make the error more clear please do let me know.

---

• treeform (orginal) [2017-12-14T17:32:19+01:00] view original

  The fact that you construct things like this thing(x:42) and call functions like this thing(x=42) is really odd. More context depended error messages could really help here.

---

• zlash (orginal) [2017-12-14T18:48:19+01:00] view original

  If there is no instance where calling a function f with f(a:b) is valid syntax, it would be nice to change the error when is found in the context of the parameters list to a function invocation.

  If it is too complicated, or not implementable, a weaker version with a help prompt could also help. Something like:

  Error: type expected. (Or maybe you are trying to call a function using named parameters in the form 'param:value' instead of 'param=value'?)

---

• refaqtor (orginal) [2017-12-14T20:24:49+01:00] view original

  thanks for the response. I had started out with

  discard os.setCurrentDir("/path/that/already/exists")
  
  Run

  as you suggested. but that returns

  Error: expression 'setCurrentDir("/path/that/already/exists")' has no type (or is ambiguous)

  in my editor and at compile time.

---

• treeform (orginal) [2017-12-14T20:55:58+01:00] view original

  os.setCurrentDir("/path/that/already/exists") does not return anything, so you don't need to discard it.

  import os
  os.setCurrentDir("/path/that/already/exists")
  Run

---

• refaqtor (orginal) [2017-12-14T21:38:40+01:00] view original

  ha! that's it! Thanks!

  the editor kept underlining the string, so, I didn't look at the discard. sometimes I can't see the forest through the trees. anyway, I'll know better how to interpret that error message! Nim error messages, by the way, have otherwise helped me tremendously to get up to speed on Nim so fast. I really appreciate all the effort that goes into Nim, and hope to contribute myself to that effort.

---

• refaqtor (orginal) [2020-04-11T15:15:46+02:00] view original

  Don't ya just hate it!? when you bang your head on a problem only to find the answer in a post to YOUR OWN FOUR YEAR OLD [solved] QUESTION !!!

  Thanks again, forum, for being here.

  Write good comments-- the coder you help may be yourself!

---

• spip (orginal) [2020-04-13T21:23:10+02:00] view original

  I would add: in any software, try to give precise and detailed error message to the user of the cause of the problem

  Since a few hours, my tests don't compile any more and the message I have is Error: type expected, but got: bool. I've tried changing the tests set, expanded manually the macros on a reduced set, etc. It's working with other tests but not this one. I know exactly the line where the problem occurs, but I can't relate it to the error message. And because that's a compilation problem, I can't call expandMacros to help the debugging...

  With compiler maximum verbosity, it seems there's a bit of context:

  
  Error: type expected, but got: bool
      result = quote do:
               ^
  Run

  That's raging...

  To solve problems, better give more details than less.

---