www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.bugs - [Issue 3960] New: Unused variable warning

reply d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960

           Summary: Unused variable warning
           Product: D
           Version: 2.041
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: enhancement
          Priority: P2
         Component: DMD
        AssignedTo: nobody puremagic.com
        ReportedBy: bearophile_hugs eml.cc



This is C program:

#include "stdio.h"
#include "stdlib.h"
int main(int argc, char** argv) {
    int x, y;
    x = (argc > 1) ? atoi(argv[1]) : 10;
    printf("%d\n", x);
    return 0;
}


Compiled with gcc v.4.4.1 with:
gcc -Wall foo.c -o foo

The compiler prints:
foo.c: In function 'main':
foo.c:4: warning: unused variable 'y'


The Intel C/C++ compiler gives a similar error (this is a mockup, but it's
equal or very close to the real one):

    int x, y;
           ^
-----------------------



using System;
public class Foo {
    public static void Main() {
        int x, y;
        x = int.Parse(Console.ReadLine());
        Console.WriteLine(x);
    }
}



prog.cs(4,16): warning CS0168: The variable `y' is declared but never used
Compilation succeeded - 1 warning(s)


This explains the warning CS0168:
http://msdn.microsoft.com/en-us/library/tf85t6e4.aspx

-----------------------

In Java all the most important editors/IDEs detect unused local variables:
JDeveloper, Eclipse, JEdit, JBuilder, BlueJ, CodeGuide, NetBeans/Sun Java
Studio Enterprise/Creator, etc.


If the most important compilers/IDEs of the most important and used languages

warning.

By itself an unused variable is not an error, but its presence is often caused
due to oversight (either you declared a variable you didn't need or you
refactored and a variable that was once needed now is no long needed). While
coding in C the GCC compiler has avoided me 2 possible bugs thanks to that
unused variable warning, because I was forgetting some part of the code.

So I believe this warning can be useful in D too.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Mar 13 2010
next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960







using System;
namespace Test {
    class Foo {
        int x;
        private void Add(int y) {
            x += y;
        }
        static void Main() {
            System.Console.WriteLine("test");
        }
    }
}



Compiled with mono:
...gmcs test.cs
test.cs(5,22): warning CS0169: The private method `Test.Foo.Add(int)' is never
used
Compilation succeeded - 1 warning(s)


In D private methods can be used by other functions in the same module, so it's
less easy to produce a warning like this.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Mar 17 2010
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960




There is more than just unused variables, there are also unused last
assignments:


void main() {
    int x;
    x = 10;
}


Here 'x' is not an unused variable, because the code does something with it,
but the code deserves a warning anyway (and one C compiler-like shows this
warning) because the last value assigned to it gets unused; and this is wasted
coding/running effort at best, or sign of a possible latent bug (just like
unused variables).

This warning isn't necessary if the variable address is assigned to a pointer
or similar things.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Aug 20 2010
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960


Jonathan M Davis <jmdavisProg gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jmdavisProg gmail.com



09:54:44 PDT ---
Okay. Giving a warning for an unused variable makes sense. However, giving a
warning for setting it somewhere that is not where it is declared doesn't make
sense. Sure, in this case, it's dumb of the programmer to have done that, but

1. The compiler should be able to optimize out such a simple case.

2. More complex cases much harder to detect, and it's not all that hard for you
to _want_ to disconnect the declaration of the variable and initializing it -
likely because of scoping issues or conditional blocks. The compiler is only

cases will likely get optimized away. Cases too complex to optimize away are
going to be much harder for the compiler to detect, and if you start going that
far, you're getting to the point where you just about should have done what
Java did and force initialization of variables rather than default initialize
them.

Not initializing a variable yourself, and the setting it later with a value
that you could have set it to at its declaration is not a coding error. It may
not be best practice, but it's not going to result in an error. So, I don't
think that it makes any sense to make it a warning.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Aug 20 2010
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960




Sorry, my mistake, I have lumped two different warnings into this enhancement
request, so please ignore comment 2

I have moved the second warning to bug 4694

Look at bug 4694 for an answer to comment 3 regarding the second warning.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Aug 20 2010
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960




See also "Using Redundancies to Find Errors", by Yichen Xie and Dawson Engler,
2002:
www.stanford.edu/~engler/p401-xie.pdf

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Sep 30 2010
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960




Two new good related flags in GCC 4.6:


New -Wunused-but-set-variable and -Wunused-but-set-parameter warnings were
added for C, C++, Objective-C and Objective-C++. These warnings diagnose
variables respective parameters which are only set in the code and never
otherwise used. Usually such variables are useless and often even the value
assigned to them is computed needlessly, sometimes expensively. The
-Wunused-but-set-variable warning is enabled by default by -Wall flag and
-Wunused-but-set-parameter by -Wall -Wextra flags.<



Some discussions in this thread:
http://www.digitalmars.com/webnews/newsgroups.php?art_group=digitalmars.D&article_id=132909

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Mar 27 2011
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960


yebblies <yebblies gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |peng2cheng2 yahoo.com



*** Issue 3163 has been marked as a duplicate of this issue. ***

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jun 10 2011
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960


Stewart Gordon <smjg iname.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|diagnostic                  |accepts-invalid
            Summary|Unused variable warning     |Unused local variables not
                   |                            |reported
           Severity|enhancement                 |normal



This is accepts-invalid, not an enhancement request.

http://www.digitalmars.com/d/1.0/function.html
"It is an error to declare a local variable that is never referred to."

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jun 10 2011
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960




PDT ---
I would point out that if it is made an error to have unused local variables,
function parameters must _not_ be included in it. That would be a major problem
in cases where overridden functions don't use all of their parameters, but they
have to have them because the base class' function does.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jun 10 2011
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960






 I would point out that if it is made an error to have unused local variables,
 function parameters must _not_ be included in it. That would be a major problem
 in cases where overridden functions don't use all of their parameters, but they
 have to have them because the base class' function does.


Don't worry, warnings against "unused local variables" and "unused argument
values" are meant to be two distinct features. Each of them needs a separate
discussion and has to stand for its own.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jun 10 2011
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960


yebblies <yebblies gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |yebblies gmail.com





 This is accepts-invalid, not an enhancement request.
 
 http://www.digitalmars.com/d/1.0/function.html
 "It is an error to declare a local variable that is never referred to."


See:
http://www.digitalmars.com/d/archives/digitalmars/D/bugs/2105.html
http://www.digitalmars.com/d/archives/digitalmars/D/Re_Wish_Variable_Not_Used_Warning_73539.html

It actually seems more likely that this is a bug in the spec, as I think Walter
is against pretty much anything that requires flow analysis.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jun 10 2011
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960








 This is accepts-invalid, not an enhancement request.
 
 http://www.digitalmars.com/d/1.0/function.html
 "It is an error to declare a local variable that is never referred to."


 
 See:
 http://www.digitalmars.com/d/archives/digitalmars/D/bugs/2105.html
 http://www.digitalmars.com/d/archives/digitalmars/D/Re_Wish_Variable_Not_Used_Warning_73539.html
 
 It actually seems more likely that this is a bug in the spec, as I think Walter
 is against pretty much anything that requires flow analysis.


In the beginning this was an enhancement request, now someone has made it a
kind of error. But in the end the situation is the same. In most cases in well
written code you don't want unused variables, and various experiments have
shown that unused experiments are often associated with bugs. On the other hand
lint tools in C that are able to spot such unused variables are seldom used,
and several of the most used compilers do spot such unused variables today. So
I conclude that I'd like the D compiler to spot them automatically for me.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jun 11 2011
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960






 It actually seems more likely that this is a bug in the spec, as I think Walter
 is against pretty much anything that requires flow analysis.


What has this issue to do with flow analysis?

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jun 11 2011
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960








 It actually seems more likely that this is a bug in the spec, as I think Walter
 is against pretty much anything that requires flow analysis.


 
 What has this issue to do with flow analysis?


You're right!

I'm still fairly sure I've seen Walter say no to this request somewhere.  If I
can find it I'll change the report to spec/enhancement.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jun 11 2011
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960






 I'm still fairly sure I've seen Walter say no to this request somewhere.  If I
 can find it I'll change the report to spec/enhancement.


This isn't an enhancement.  That the compiler doesn't behave according to the
spec is an outright bug.

It might be a case of removing the statement from the spec.  But
- it won't stop people wanting the compiler to warn of an unused local
variable, so this would remain valid albeit as an enhancement request
- any clue of why Walter may have changed his mind about this?

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jun 11 2011
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960








 I'm still fairly sure I've seen Walter say no to this request somewhere.  If I
 can find it I'll change the report to spec/enhancement.


 
 This isn't an enhancement.  That the compiler doesn't behave according to the
 spec is an outright bug.
 
 It might be a case of removing the statement from the spec.  But
 - it won't stop people wanting the compiler to warn of an unused local
 variable, so this would remain valid albeit as an enhancement request
 - any clue of why Walter may have changed his mind about this?


All I can find is
http://www.digitalmars.com/d/archives/digitalmars/D/Unused_variables_better_as_error_or_warning_115751.html#N115794
and
http://www.digitalmars.com/d/archives/c++/command-line/282.html
(from 2003!)

If the spec is wrong, then this _is_ a valid enhancement _and_ a valid spec
bug.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jun 11 2011
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960


Marco Leise <Marco.Leise gmx.de> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |Marco.Leise gmx.de



---


 I would point out that if it is made an error to have unused local variables,
 function parameters must _not_ be included in it. That would be a major problem
 in cases where overridden functions don't use all of their parameters, but they
 have to have them because the base class' function does.


In languages where you cannot omit the name of the parameter it is a problem.
In D you just leave the parameter name away and this problem is solved + it
becomes evident from looking at the method, that it doesn't use a certain
parameter. That would make a nice coding standard as well.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Feb 02 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960


timon.gehr gmx.ch changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |timon.gehr gmx.ch



Whether or not a parameter is used is unrelated to whether or not it has a
name.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Feb 05 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960




---


 Whether or not a parameter is used is unrelated to whether or not it has a
 name.


You can't use an unnamed parameter, or can you?
What I wanted to hint at is this:

interface I {
  void foo(int a, int b);
}

class C {
  int sum_a;
  void foo(int a, int b) {
    sum_a += a;
  }
}

This is a typical situation I face in Java code. The IDE (Eclipse in this case)
would warn me, that I don't use parameter 'b'. Now my code is perfectly fine
and my implementation of foo() doesn't depend on the value of 'b' to do its
job.
The better solution is to write:

  void foo(int a, int) {
    sum_a += a;
  }

Now the interface is happy, as is the compiler check, because we made it clear,
that we intended to ignore 'b'.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Feb 09 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960






 Now the interface is happy, as is the compiler check, because we 
 made it clear, that we intended to ignore 'b'.


Indeed.  Some C++ compilers work on the principle that if you named a
parameter, you intended to use it, and so issue a warning if you haven't used
it.

But this bug report is about unused local variables, not unused function
parameters.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Feb 09 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960




---
You are right. Since local variables and parameters are related, it could be
that the person fixing this also thinks about the situation for function
parameters. A new bug report or enhancement request would cause twice the work
in this case.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Feb 09 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960




PST ---
I know that Walter is against having warnings or errors for unused parameters
(he has _definitely_ said as much on the newsgroup), but he may feel
differently about local variables. The spec specifically says that unused
_local_ variables are an error. So, Walter probably views those differently,
which makes some sense - particularly since they aren't affected by overriding
and the like.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Feb 09 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960








 Whether or not a parameter is used is unrelated to whether or not it has a
 name.


 
 You can't use an unnamed parameter, or can you?
 


I can:
import std.stdio;
double foo(double){return _param_0;}
void main(){writeln(foo(2));}

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Feb 18 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960




I have found an interesting information.

In Go language an unused variable is an _error_:


package main
func main() {
  x := 10;
}


prog.go:3: x declared and not used

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Mar 08 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960




See also Issue 2197  and  Issue 4694

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Apr 24 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960




PDT ---

warnings or errors for unused variables. Such would needlessly make writing
template constraints harder.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Apr 26 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960




This discussion is relevant to both this and issue 7989, so I'm continuing it
on digitalmars.D.learn under the existing "Docs: Section on local variables"
thread.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Apr 26 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960







 warnings or errors for unused variables. Such would needlessly make writing
 template constraints harder.


The fact that in some uncommon situations you want to define a variable and not
use it, can't justify the lack of a compile reaction to unused variables. There
are solutions to that problem, Steven Schveighoffer suggests something like:

pragma(used) int x;

Or:
 used int x;

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Apr 26 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960






 There are solutions to that problem, Steven Schveighoffer suggests 
 something like:
 
 pragma(used) int x;
 
 Or:
  used int x;


Does this gain anything over what I suggested in issue 7989 comment 3?

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Apr 26 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960


rswhite4 googlemail.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rswhite4 googlemail.com



I wrote a short programm which detect and list unused variables.
I'm sure it isn't perfect but it passed most of my test cases and IMO 
something like this should be integrated into the D compiler.
Maybe automatically if you use the -w or -wi compiler flag switch.

Code: http://dpaste.dzfl.pl/c5f1d2ba
Some of my testcases: http://dpaste.dzfl.pl/f74da85e

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jul 30 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960







 Code: http://dpaste.dzfl.pl/c5f1d2ba
 Some of my testcases: http://dpaste.dzfl.pl/f74da85e


Paste sites eventually lose their content. That's why I have said to *attach*
your stuff in Bugzilla.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jul 30 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960


rswhite4 googlemail.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|nobody puremagic.com        |rswhite4 googlemail.com



Created an attachment (id=1128)
Unused variables detector code

Forget to upload the code of my unused variables detector.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jul 30 2012
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960


Stewart Gordon <smjg iname.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------

          mime type|                            |


-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Aug 23 2012
prev sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3960




A related warning is for "unused local alias".

A bug in GCC, libstdc++/33084, in <valarrray> function with this body:


typedef _BinClos<_Name, _Constant, _ValArray, _Tp, _Tp> _Closure;

typedef typename __fun<_Name, _Tp>::result_type _Rt;

return _Expr<_Closure, _Tp>(_Closure(__t, __v));


There is a typo: _Rt  ->  _Tp

In GCC 4.7, the -Wunused-local-typedefs warning detects such sort of suspect
"unused" typedef.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Oct 21 2012