Print debugging statement: Difference between revisions

From Rosetta Code
Content added Content deleted
Line 237: Line 237:
end function
end function


-- NOTE: following four routines must all use the exact same nesting level.
-- NOTE: following five routines must all use the exact same nesting level.


global function debug_info()
global function debug_info()
Line 245: Line 245:
global function debug_line()
global function debug_line()
return _debug_info()[E_LINE]
return _debug_info()[E_LINE]
end function

global function debug_rtn()
return _debug_info()[E_NAME]
end function
end function



Revision as of 11:58, 5 September 2019

Print debugging statement is a draft programming task. It is not yet considered ready to be promoted as a complete task, for reasons that should be found in its talk page.

From Wikipedia:

Print debugging (or tracing) is the act of watching (live or recorded) trace statements, or print statements, that indicate the flow of execution of a process. This is sometimes called printf debugging, due to the use of the printf function in C.

Task
  • Show the print debugging statements in the language.
  • Demonstrate their ability to track provenance by displaying information about source code (e.g., code fragment, line and column number).



C

C doesn't have a built-in print debugging statement. However, it can be defined by users as a macro.

<lang C>#include <stdio.h>

  1. define DEBUG_INT(x) printf( #x " at line %d\nresult: %d\n\n", __LINE__, x)

int add(int x, int y) {

 int result = x + y;
 DEBUG_INT(x);
 DEBUG_INT(y);
 DEBUG_INT(result);
 DEBUG_INT(result+1);
 return result;

}

int main() {

 add(2, 7);
 return 0;

}</lang>

Output:
x at line 7
result: 2

y at line 8
result: 7

result at line 9
result: 9

result+1 at line 10
result: 10

Go

Go doesn't have a built-in print debugging statement as such. Nor does it have macros.

However, as the following example shows, it is easy enough to mimic a C-like approach by writing a short 'debug' function which can show the value of an expression and its type at the appropriate line number in the program's source code.

Note that a label for the expression (whether it's a simple variable or not) must be passed to the 'debug' function as there is no way to deduce it otherwise. <lang go>package main

import (

   "fmt"
   "runtime"

)

type point struct {

   x, y float64

}

func add(x, y int) int {

   result := x + y
   debug("x", x)
   debug("y", y)
   debug("result", result)
   debug("result+1", result+1)
   return result

}

func debug(s string, x interface{}) {

   _, _, lineNo, _ := runtime.Caller(1)
   fmt.Printf("%q at line %d type '%T'\nvalue: %#v\n\n", s, lineNo, x, x)

}

func main() {

   add(2, 7)
   b := true
   debug("b", b)
   s := "Hello"
   debug("s", s)
   p := point{2, 3}
   debug("p", p)
   q := &p
   debug("q", q)

}</lang>

Output:
"x" at line 14 type 'int'
value: 2

"y" at line 15 type 'int'
value: 7

"result" at line 16 type 'int'
value: 9

"result+1" at line 17 type 'int'
value: 10

"b" at line 29 type 'bool'
value: true

"s" at line 31 type 'string'
value: "Hello"

"p" at line 33 type 'main.point'
value: main.point{x:2, y:3}

"q" at line 35 type '*main.point'
value: &main.point{x:2, y:3}

Perl 6

Works with: Rakudo version 2019.07.1

There isn't anything built-in to do this in Rakudo Perl 6, though it's pretty easy to cobble something together piggybacking off of the exception system. It would probably be better to instantiate a specific "DEBUG" exception to avoid interfering with other user instantiated ad-hoc exceptions, but for a quick-and-dirty demo, this should suffice.

This example will report any payload contents passed to the exception. If you want specific information, it will need to be passed in, though some of it may be determinable through introspection. Reports the file name and line number where the "debug" call originated and unwinds the call stack to trace through the subroutine calls leading up to it. Will follow the call chain into included files and modules, though calls to the CORE setting and dispatcher are filtered out here to reduce noise.

Comments with the files line numbers are added here to make it easier to match up the debug output with the file. Typically you would be editing the file in an editor that provides line numbering so that wouldn't be necessary/helpful.

<lang perl6>my &pdb = ¨

CATCH {

   when X::AdHoc {
       my @frames = .backtrace[*];
       say .payload;
       for @frames {
           # Filter out exception handing and dispatcher frames
           next if .file.contains: 'SETTING' or .subname.chars < 1;
           printf "%sfrom file: %s,%s line: %s\n",
             (' ' x $++), .file,
             (my $s = .subname) eq '<unit>' ??  !! " sub: $s,", .line;
       }
       say ;
       .resume;
   }
   default {}

}

    1. Testing / demonstration
  1. helper subs #line 22

sub alpha ($a) { #line 23

   pdb ('a =>', $a + 3);    #line 24
   pdb 'string';            #line 25
   beta(7);                 #line 26

} #line 27 sub beta ($b) { pdb $b } #line 28 sub gamma ($c) { beta $c } #line 29 sub delta ($d) { gamma $d } #line 30

                            #line 31

my $a = 10; #line 32 pdb (.VAR.name, $_) with $a; #line 33 alpha($a); #line 34 delta("Δ"); #line 35 .&beta for ^3; #line 36</lang>

Output:
($a 10)
from file: debug.p6, line: 33

(a => 13)
from file: debug.p6, sub: alpha, line: 24
 from file: debug.p6, line: 34

string
from file: debug.p6, sub: alpha, line: 25
 from file: debug.p6, line: 34

7
from file: debug.p6, sub: beta, line: 28
 from file: debug.p6, sub: alpha, line: 26
  from file: debug.p6, line: 34

Δ
from file: debug.p6, sub: beta, line: 28
 from file: debug.p6, sub: gamma, line: 29
  from file: debug.p6, sub: delta, line: 30
   from file: debug.p6, line: 35

0
from file: debug.p6, sub: beta, line: 28
 from file: debug.p6, line: 36

1
from file: debug.p6, sub: beta, line: 28
 from file: debug.p6, line: 36

2
from file: debug.p6, sub: beta, line: 28
 from file: debug.p6, line: 36

Phix

The ? statement is used as a quick shorthand to dump variable contents and expression results.
For provenance, the following was added to builtins/reflections.e (for now/not an autoinclude, 0.8.1+), and throw() was also tweaked to convert supplied addresses, which it previously did not. It proved to be quite an easy enhancement to the language, albeit as yet undocumented. <lang Phix>function _debug_info() -- use throw to convert a return address and routine number -- from the call stack into a proper line number, etc. -- (private, not called direct/from outside this file) integer rtn atom ret_addr

   #ilASM{
       [32]
           mov edx,[ebp+20]    -- prev_ebp
           mov eax,[edx+28]    -- return address
           mov edx,[edx+20]    -- prev_ebp
           lea edi,[ret_addr]
           call :%pStoreMint
           mov eax,[edx+8]     -- calling routine no
           mov [rtn],eax
       [64]
           mov rdx,[rbp+40]    -- prev_ebp
           mov rax,[rdx+56]    -- return address
           mov rdx,[rdx+40]    -- prev_ebp
           lea rdi,[ret_addr]
           call :%pStoreMint
           mov rax,[rdx+16]    -- calling routine no
           mov [rtn],rax
       []
         }
   try
       throw({1,ret_addr-1,-1,rtn,-1,-1,-1})
   catch e
       return e
   end try

end function

-- NOTE: following five routines must all use the exact same nesting level.

global function debug_info()

   return _debug_info()

end function

global function debug_line()

   return _debug_info()[E_LINE]

end function

global function debug_rtn()

   return _debug_info()[E_NAME]

end function

global function debug_file()

   return _debug_info()[E_FILE]

end function

global function debug_path()

   return _debug_info()[E_PATH]

end function</lang> This can be used as follows (0.8.1+) <lang Phix>include builtins/reflections.e

?debug_info() ?debug_line() procedure test()

   ?debug_info()
   ?debug_line()
   ?debug_file()
   printf(1,"This is line %d in file %s\n",{debug_line(),debug_file()})

end procedure test()</lang>

Output:
{1,7544919,3,21,"-1","test.exw","C:\\Program Files (x86)\\Phix\\"}
4
{1,7545341,6,1034,"test","test.exw","C:\\Program Files (x86)\\Phix\\"}
7
"test.exw"
This is line 9 in file test.exw

See the throw() documentation for more details, especially regarding the debug_info() results.
There is no routine name for the first debug_info() call, so you get "-1" in [E_NAME].
The line numbers 3, 4, 6, 7, and 9 are returned for five of the seven calls.

Pyret

Pyret has the spy expression. The expression can print the value of an identifier, using the identifier itself as a label if it's not already given. It could also print the value of an arbitrary expression, but it needs an explicit label in this case.

<lang pyret>fun add(x, y):

 result = x + y
 spy "in add": 
   x,
   y,
   result,
   result-plus-one: result + 1
 end
 result

end

add(2, 7)</lang>

Output:
Spying "in add" (at file:///spies.arr:3:2-8:5)
  x: 2
  y: 7
  result: 9
  result-plus-one: 10

9

Racket

Racket doesn't have a built-in print debugging statement. However, it can be defined by users as a macro.

<lang racket>#lang racket

(require syntax/parse/define)

(define (debug:core line col code val #:label [label #f])

 ;; if label exists, use it instead of the code fragment
 (printf "~a at line ~a column ~a\n" (or label code) line col)
 (printf "result: ~a\n\n" val)
 ;; return the value itself, so that we can wrap macro around an expression 
 ;; without restructuring any code
 val)

(define-simple-macro (debug <x> option ...)

 #:with line (datum->syntax this-syntax (syntax-line #'<x>))
 #:with col (datum->syntax this-syntax (syntax-column #'<x>))
 (debug:core line col (quote <x>) <x> option ...))

(define (add x y)

 (define result (+ x y))
 (debug x)
 (debug y)
 (debug (if #t (+ x y) (error 'impossible)))
 (debug (add1 result) #:label "result plus one")
 (debug result))

(add 2 7)</lang>

Output:
x at line 20 column 9
result: 2

y at line 21 column 9
result: 7

(if #t (+ x y) (error 'impossible)) at line 22 column 9
result: 9

result plus one at line 23 column 9
result: 10

result at line 24 column 9
result: 9

9

REXX

There are other options for the REXX's   trace   instruction,   but the   i   is the most informative and
shows intermediate results within a REXX statement as it's being evaluated.

The first number   (for the   trace   output)   is the line number for the REXX program.
Blank lines are not   traced.

This particular output is from the Regina REXX interpreter. <lang rexx>/*REXX program to demonstrate debugging (TRACE) information while executing a program*/ /*────────────────────────────────────────────── (below) the I is for information. */ trace i parse arg maxDiv . if maxDiv== | maxDiv=="," then maxDiv= 1000 /*obtain optional argument from the CL.*/ say 'maximum random divisor is:' maxDiv /*display the max divisor being used. */ total= 0

        do j=1  to 100
        total= total + j/random(maxDiv)
        end   /*j*/

say 'total=' total /*stick a fork in it, we're all done. */</lang>

output   when using the input of:     9
     4 *-* parse arg maxDiv .
       >>>   "9"
       >.>   ""
     5 *-* if maxDiv=='' | maxDiv==","  then maxDiv= 1000   /*obtain optional argument from the CL.*/
       >V>   "9"
       >L>   ","
       >O>   "0"
       >V>   "9"
       >L>   ""
       >O>   "0"
       >U>   "0"
     6 *-* say 'maximum random divisor is:'  maxDiv         /*display the max divisor being used.  */
       >L>   "maximum random divisor is:"
       >V>   "9"
       >O>   "maximum random divisor is: 9"
maximum random divisor is: 9
     7 *-* total= 0
       >L>   "0"
     9 *-* do j=1  to 100
       >L>   "1"
       >L>   "100"
       >V>   "1"
    10 *-*  total= total + j/random(maxDiv)
       >V>    "0"
       >V>    "1"
       >V>    "9"
       >F>    "3"
       >O>    "0.333333333"
       >O>    "0.333333333"
    11 *-* end   /*j*/
     9 *-* do j=1  to 100
       >V>   "1"
       >V>   "2"
    10 *-*  total= total + j/random(maxDiv)
       >V>    "0.333333333"
       >V>    "2"
       >V>    "9"
       >F>    "0"
    10 +++    total= total + j/random(maxDiv)
Error 42 running "c:\debuggin.rex", line 10: Arithmetic overflow/underflow
Error 42.3: Arithmetic overflow; divisor must not be zero

zkl

Print debugging is similar to C. The _debug_ keyword conditionally compiles code (ie the debug code isn't compiled unless debugging is turned on). <lang zkl>fcn ds(line=__LINE__){

  println("This is line %d of file %s compiled on %s"
          .fmt(line,__FILE__,__DATE__));

}(); _debug_{

  ds(__LINE__); println("Debug level is ",__DEBUG__);
  vm.stackTrace().println();

}</lang>

Output:
Run with debugging turned off:

$ zkl rs
This is line 39 of file rs.zkl compiled on 2019-08-28

Run with debugging turned on:
Due to some brain deadness, we need to set the debug level (-d), compile the 
file (-c) then run it and quit out of the REPL:

$ zkl -dc rs --run --quit
Compiled Class(rs)  (0.0 seconds, ??? lines/sec)
This is line 44 of file rs.zkl compiled on 2019-08-28
This is line 49 of file rs.zkl compiled on 2019-08-28
Debug level is 1
Stack trace for VM#1 ():
   rs.__constructor@stackTrace addr:25  args(0) reg(0) 
   startup.__constructor@__constructor addr:1767  args(0) reg(22) R