Why does this code work?

  • Thread starter Thread starter Daniel Billingsley
  • Start date Start date
D

Daniel Billingsley

Or seem to anyway.
=====
string abc;
if ((abc=getString()) != "")
{
MessageBox.Show(abc);
}

private string getString()
{
return "something";
}
=====
I get the MessageBox displayed. If I change != to == there's no MessageBox.

What is the != comparing, and how?
 
Daniel Billingsley said:
Or seem to anyway.
=====
string abc;
if ((abc=getString()) != "")
{
MessageBox.Show(abc);
}
Click to expand...

As your parenthesis indicate, the first thing that happens is:
(abc=getString())
so, abc gets the value of "something"

Which is then tested to be != to "", which is true. "something" != "" so the messagebox shows. Why do you feel there it should be
otherwise?
 
Hi,

Since abc is assigned to the value returned by getString() and then that
value is compared to "" (condition validates to true), MessageBox is shown.
In case of ==, the condition validates to false and no messagebox.

Or seem to anyway.
=====
string abc;
if ((abc=getString()) != "")
{
MessageBox.Show(abc);
}

private string getString()
{
return "something";
}
=====
I get the MessageBox displayed. If I change != to == there's no MessageBox.

What is the != comparing, and how?
 
Daniel Billingsley said:
Or seem to anyway.
=====
string abc;
if ((abc=getString()) != "")
{
MessageBox.Show(abc);
}

private string getString()
{
return "something";
}
=====
I get the MessageBox displayed. If I change != to == there's no MessageBox.

What is the != comparing, and how?
Click to expand...

It's comparing "" to the value of the expression (abc=getString())
which is the value of abc after the assignment, which is "something".
It's the same way that the common:

while ( (line=streamReader.ReadLine()) != null)

or

while ( (read=stream.Read(buffer, 0, buffer.Length)) > 0)

work.
 
Or seem to anyway.
=====
string abc;
if ((abc=getString()) != "")
{
MessageBox.Show(abc);
}

private string getString()
{
return "something";
}
=====
I get the MessageBox displayed. If I change != to == there's no
MessageBox.

What is the != comparing, and how?
Click to expand...

Interesting question. Let's see what the compiler is up to by using ILDASM
to look at the generated IL code. I've put the code in the click event of
a button.


..method private hidebysig instance void button1_Click(object sender,
class [mscorlib]
System.EventArgs e) cil managed
{
// Code size 32 (0x20)
.maxstack 2
.locals init ([0] string abc)

//push the this pointer onto the stack
IL_0000: ldarg.0
//Stack now looks like:
// this


//load the return value of getString onto the stack
IL_0001: call instance string WindowsApplication5.Form1::getString
()
//Stack now looks like:
// this, "Something"

//Copy the top most value on the stack
IL_0006: dup
//Stack now looks like:
// this, "Something", "Something"

//Store the topmost value on the stack to
//the first local variable (abc in this case)
IL_0007: stloc.0
//Stack now looks like:
// this, "Something"

//Load an empty string onto the stack
IL_0008: ldstr ""
//Stack now looks like:
// this, "Something", ""

//Test the top two values on the stack for
//Inequality
IL_000d: call bool [mscorlib]System.String::op_Inequality
string,string)
//Stack now looks like:
// this

//If False then jump to IL_001f
IL_0012: brfalse.s IL_001f

//Since the comparison of the top two stack values
//returns true you get the message box
//Load the string onto the stack
IL_0014: ldstr "Hello Something"
//Stack now looks like:
// this, "Hello Something"

//Show the message box
IL_0019: call valuetype [System.Windows.Forms]
System.Windows.Forms.DialogResult [System.Windows.Forms]
System.Windows.Forms.MessageBox::Show(string)
//Stack now looks like:
// this

//pop the this pointer off the stack
IL_001e: pop

//return
IL_001f: ret
} // end of method Form1::button1_Click

So, now you can see what is happening behind the scenes which should help
to explain what was going on in that expression and why it works.

Dave
 
As your parenthesis indicate, the first thing that happens is:
(abc=getString())
so, abc gets the value of "something"

Which is then tested to be != to "", which is true. "something" != ""
so the messagebox shows. Why do you feel there it should be
otherwise?
Click to expand...

Perhaps because the code reads:

If you assign the return value of getString() to abc and compare that
_operation_ to "" testing for inequality then...

(abc = getString()) has no return value and fails by throwing an exception
if something goes wrong; so, as the person asked, what is != comparing
with?

See my other reply for the answer.

Dave
 
(abc = getString()) has no return value and fails by throwing an exception
if something goes wrong; so, as the person asked, what is != comparing
with?
Click to expand...

What do you mean it has no return value... it most certainly does. Returns the value of the string assigned. The assignment
operator returns the value assigned (or reference, whatever you happen to be dealing with).
It is what allows:
string a = "a";
string b, c;
b = c = a;

Also, the same principle as something like this would work:
StreamReader sr = .....
string line;
while ( (line = sr.ReadLine()) != null)
{
....
}

If the assignment operator did not return a value, neither of the above would be possible.
 
What do you mean it has no return value... it most certainly does.
Returns the value of the string assigned. The assignment operator
returns the value assigned (or reference, whatever you happen to be
dealing with). It is what allows:
string a = "a";
string b, c;
b = c = a;

Also, the same principle as something like this would work:
StreamReader sr = .....
string line;
while ( (line = sr.ReadLine()) != null)
{
...
}

If the assignment operator did not return a value, neither of the
above would be possible.
Click to expand...

Not the *operator*. The actual operation that's in parens.

(abc = getString()); is a stand alone operation. abc gets the value of
getString() but the code does not read:

abc = getString();
if(abc != "")

In the example, the left operand of the != operator is an expression.
There's no extra "=" operator as in your example so that something like
this would be obvious:

(a = (b = c) != "")

which is effectively the same as
b = c;
a = b;
a != "";

What we have is:

b = c;
? != "";
 
David Totzke (.NET/C# MVP) said:
Perhaps because the code reads:

If you assign the return value of getString() to abc and compare that
_operation_ to "" testing for inequality then...

(abc = getString()) has no return value and fails by throwing an exception
if something goes wrong; so, as the person asked, what is != comparing
with?
Click to expand...

No, the expression (abc=getString()) *does* have a value (not a return
value, but a value).

From section 14.13.1 of the ECMA C# spec:

<quote>
http://www.pobox.com/~skeet/csharp/faq/#db.parameters
</quote>

That value is what != is comparing with.
 
David Totzke (.NET/C# MVP) said:
Not the *operator*. The actual operation that's in parens.

(abc = getString()); is a stand alone operation. abc gets the value of
getString() but the code does not read:

abc = getString();
if(abc != "")
Click to expand...

Although it doesn't read that way, it is semantically equivalent to
that.
In the example, the left operand of the != operator is an expression.
Click to expand...

Yes, and the value of the expression is the value assigned to the left-
hand side of the assignment operator, as specified in the C# language
specification.
 
Hi,
inline

David Totzke (.NET/C# MVP) said:
Not the *operator*. The actual operation that's in parens.

(abc = getString()); is a stand alone operation. abc gets the value of
getString() but the code does not read:

abc = getString();
if(abc != "")

In the example, the left operand of the != operator is an expression.
There's no extra "=" operator as in your example so that something like
this would be obvious:

(a = (b = c) != "")

which is effectively the same as
b = c;
a = b;
a != "";
Click to expand...

No, it's not.

(a = (b = c) != "") doesn't compile, and it's effectively the same as:

b = c;
a = (b!=""); // ERROR, assigning bool to string

What we have is:

b = c;
? != "";
Click to expand...

Wrong. (b=c) has the value of b after c is assigned to it.


Greetings
 
Yes, and the value of the expression is the value assigned to the left-
hand side of the assignment operator, as specified in the C# language
specification.
Click to expand...

The original responses to the question were basically things like this:
"What did you expect?"

I took a shot and showed under the covers why it does work. Not being one
for reading specs the statement's functionality does seem counter
intuitive. The spec makes it plain. Your answer now is a good one and
I've learned something from the exercise.

Someone pointed out one of the flaws in my analysis in that
(a = (b = c) != "") doesn't compile because (b = c) evaluates to a bool and
we get the old implicit type conversion error. So my next question would
be:

Why does...oh CRAP! I just saw my mistake. It's trying to assign the
value of (b = c) != "" to a. A bool and a string.

I originally went astray because I know that at some point I have hosed
myself by using the assignment operator in an if condition instead of the
comparison and it was never false because the assignment always worked.
This if(b = c){} won't compile in C# but I know I did it somewhere. Nearly
went insane until I spotted it. C++ maybe? Can anybody tell me?

Thanks for the help.

Cheers,
Dave
 
David Totzke (.NET/C# MVP) said:
This if(b = c){} won't compile in C# but I know I did it somewhere. Nearly
went insane until I spotted it. C++ maybe? Can anybody tell me?
Click to expand...

Yes, that works in C/C++, which is why you often see people writing

if (5==x)

instead of

if (x==5)

in case you make a typo and turn it into an assignment instead of a
comparison. In C# this is not needed (unless you're comparing boolean
constants, and there I usually just write if(x) or if(!x)) and makes
the code less readable, IMO.
 
Hi,
I originally went astray because I know that at some point I have hosed
myself by using the assignment operator in an if condition instead of the
comparison and it was never false because the assignment always worked.
This if(b = c){} won't compile in C# but I know I did it somewhere. Nearly
went insane until I spotted it. C++ maybe? Can anybody tell me?
Click to expand...

Yes, this may compile in c++, but it has the same effect as in c#. The
difference is that C# won't do an implicit cast.

<Quote from C++ iso standard 5.17>
There are serveral assignment operators, all of which group right-to-left.
All require a modifiable lvalue as their left operand and the type of an
assignment expression is that of its left operand. The result of assignment
is the value stored in the left operand after the assignment has taken
place; the result is an lvalue.
</Quote>

<Quote from C++ iso standard 6.4>
The value of a condition that is an expression is the value of the
expression, implicitly converted to bool ...
</Quote>

Rules for casting a number to bool are (true!=0,false=0).

int b;
if ( b = 5 )
{
// true, not because assignment succeeds, but because (b=5) is 5 which
is != 0
}



HTH,
greetings
 
Yes, that works in C/C++, which is why you often see people writing

if (5==x)

instead of

if (x==5)
Click to expand...

Jon,
Actually you won't see this often. It is considered by C programers
as a juvenile coding practice used by people that don't proof read
their own code.

Oz
 
Actually you won't see this often. It is considered by C programers
as a juvenile coding practice used by people that don't proof read
their own code.
Click to expand...

Well, I'm sure it's considered juvenile by some people, but definitely
not by others - and I've seen it reasonably often in C# code (where
it's useless in most cases)...
 
Well, I'm sure it's considered juvenile by some people, but definitely
not by others - and I've seen it reasonably often in C# code (where
it's useless in most cases)...
Click to expand...

Exactly my point (the useless comment). The "often" comment of
mine was with regards to C programmers, not C# programmers BTW.

And I've never seen it in used in any MSDN example.

BTW...thanks for the great article on threading. I don't know how
you find the time but I am glad you do/did!

Oz
 
Back
Top