Why does the stacktrace stops short of the error?

Questions & Answers
1

I've created a minimal example for a nasty error I had:

#lang racket

(require "foo.rkt")

(module+ main
  (run))

(define (run)
  (error-fun 7))

and foo.rkt:

#lang racket

(provide error-fun)

(struct foo (num))

(define (error-fun foo)
  (foo 8))

When running this with racket main.rkt I get this error:

application: not a procedure;
 expected a procedure that can be applied to arguments
  given: 7
  context...:
   body of (submod "/home/ingmar/projects/racket-stacktrace/main.rkt" main)

Mind, I know what the error is: because error-fun has a parameter named foo, the line below can't create a struct of type foo (which was the intention). I would've to rename either the struct or the parameter.

The problem I have is that the error message is not in the least helpful to find the error. Mind, this is a minimal example; I've had a similar error in a much larger codebase.

Specifically:

  • The file with the error isn't mentioned at all (foo.rkt).
  • The function with the error isn't mentioned either (error-fun).
  • The offending line in the code isn't quoted nor the line number is displayed.

My question is, is there a better way of finding an error like this? Is there a coding style I should follow to get better error messages?

2

Someone else may be able to explain the lower level details as to why this is, but what you want is to compile your code with the errortrace module. See:

So running racket -l errortrace -t main.rkt will show the line triggering the error:

application: not a procedure;
 expected a procedure that can be applied to arguments
  given: 7
  errortrace...:
   /home/jonathan/foo.rkt:8:2: (foo 8)
  context...:
   body of (submod "/home/jonathan/main.rkt" main)

Drracket and emacs(with racket-mode) both have built-in methods to run your code with errortrace, which makes it easy if you use those.

1 Like