preprocessor q: impossible macro?

[Eric]

Hello all,
I've got this line of given code (cannot change this; wizard-generated, but
value may change someday):
#define IDB_BUTTON 225

Somewhere in the code I found / need this:
.....somefuncti on("#225")....

This string in the function call is the same number like in the define, but
the constant wasn't used. Because for future changes and for readability I
wanted to create this string with a preprocessor macro like this:
.....somefuncti on(MYMACRO(IDB_ BUTTON))....

I didn't get such a macro to work. I tried:
#define MYMACRO(num) "#" #num
but this only yields to
.....somefuncti on("#" "IDB_BUTTON").. ..
I also tried with the token-pasting operator (##) etc.

Is this possible at all?
Eric

[Victor Bazarov]

Re: preprocessor q: impossible macro?

Eric wrote:[color=blue]
> I've got this line of given code (cannot change this; wizard-generated, but
> value may change someday):
> #define IDB_BUTTON 225
>
> Somewhere in the code I found / need this:
> ....somefunctio n("#225")....
>
> This string in the function call is the same number like in the define, but
> the constant wasn't used. Because for future changes and for readability I
> wanted to create this string with a preprocessor macro like this:
> ....somefunctio n(MYMACRO(IDB_B UTTON))....
>
> I didn't get such a macro to work. I tried:
> #define MYMACRO(num) "#" #num
> but this only yields to
> ....somefunctio n("#" "IDB_BUTTON").. ..
> I also tried with the token-pasting operator (##) etc.
>
> Is this possible at all?[/color]

Yes, you need an indirect "stringizin g" macro for that.

#define STR(x) #x
#define MYMACRO(num) "#" ## STR(num)

#define IDB_BUTTON 225

#include <stdio.h>

int main(void) {
printf("Got %s\n", MYMACRO(IDB_BUT TON));
}

V

[Charles Mills]

Re: preprocessor q: impossible macro?

Victor Bazarov wrote:[color=blue]
> Eric wrote:[color=green]
> > I've got this line of given code (cannot change this; wizard-generated, but
> > value may change someday):
> > #define IDB_BUTTON 225
> >
> > Somewhere in the code I found / need this:
> > ....somefunctio n("#225")....
> >
> > This string in the function call is the same number like in the define, but
> > the constant wasn't used. Because for future changes and for readability I
> > wanted to create this string with a preprocessor macro like this:
> > ....somefunctio n(MYMACRO(IDB_B UTTON))....
> >
> > I didn't get such a macro to work. I tried:
> > #define MYMACRO(num) "#" #num
> > but this only yields to
> > ....somefunctio n("#" "IDB_BUTTON").. ..
> > I also tried with the token-pasting operator (##) etc.
> >
> > Is this possible at all?[/color]
>
> Yes, you need an indirect "stringizin g" macro for that.
>
> #define STR(x) #x
> #define MYMACRO(num) "#" ## STR(num)
>
> #define IDB_BUTTON 225
>
> #include <stdio.h>
>
> int main(void) {
> printf("Got %s\n", MYMACRO(IDB_BUT TON));
> }
>
> V[/color]

The above solution does not compile. Token pasting is not necessary
since adjacent string literals are pasted.

#define STR(x) #x
#define MYMACRO(num) "#" STR(num)

#define IDB_BUTTON 225

#include <stdio.h>

int main(void) {
printf("Got %s\n", MYMACRO(IDB_BUT TON));
}

-Charlie

[Victor Bazarov]

Re: preprocessor q: impossible macro?

Charles Mills wrote:[color=blue]
> Victor Bazarov wrote:
> [..][color=green]
>>
>> #define STR(x) #x
>> #define MYMACRO(num) "#" ## STR(num)
>>
>> #define IDB_BUTTON 225
>>
>> #include <stdio.h>
>>
>> int main(void) {
>> printf("Got %s\n", MYMACRO(IDB_BUT TON));
>> }
>>
>>V[/color]
>
>
> The above solution does not compile. [..][/color]

On what compiler? What is the error message?

V

[Charles Mills]

Re: preprocessor q: impossible macro?



Victor Bazarov wrote:[color=blue]
> Charles Mills wrote:[color=green]
> > Victor Bazarov wrote:
> > [..][color=darkred]
> >>
> >> #define STR(x) #x
> >> #define MYMACRO(num) "#" ## STR(num)
> >>
> >> #define IDB_BUTTON 225
> >>
> >> #include <stdio.h>
> >>
> >> int main(void) {
> >> printf("Got %s\n", MYMACRO(IDB_BUT TON));
> >> }
> >>
> >>V[/color]
> >
> >
> > The above solution does not compile. [..][/color]
>
> On what compiler? What is the error message?
>
> V[/color]

gcc
pasting ""#"" and "STR" does not give a valid preprocessing token

The standard says that token pasting must result in a valid C token.

-Charlie

[Victor Bazarov]

Re: preprocessor q: impossible macro?

Charles Mills wrote:[color=blue]
> Victor Bazarov wrote:
>[color=green]
>>Charles Mills wrote:
>>[color=darkred]
>>>Victor Bazarov wrote:
>>>[..]
>>>
>>>> #define STR(x) #x
>>>> #define MYMACRO(num) "#" ## STR(num)
>>>>
>>>> #define IDB_BUTTON 225
>>>>
>>>> #include <stdio.h>
>>>>
>>>> int main(void) {
>>>> printf("Got %s\n", MYMACRO(IDB_BUT TON));
>>>> }
>>>>
>>>>V
>>>
>>>
>>>The above solution does not compile. [..][/color]
>>
>>On what compiler? What is the error message?
>>
>>V[/color]
>
>
> gcc
> pasting ""#"" and "STR" does not give a valid preprocessing token
>
> The standard says that token pasting must result in a valid C token.[/color]

I don't have a copy of the _old_ C standard handy, but the C++ Standard
has the sub-clause 16.3.1 (and C99 has 6.10.3.1), which essentially says
that operators # and ## are only acted upon after *all* substitutions have
taken place. So, your version of gcc is *buggy* if it doesn't compile the
code I posted, it ought to rescan the substitution (according to 16.3.4)
until no macro is left unsubstituted.

IOW, the result of 'MYMACRO(IDB_BU TTON)' yields

"#" ## STR(225)

which in turn, after rescanning yields

"#" ## #225

which then yields

"#" ## "225"

And two tokens are concatenated. Relying on the catenation of strings
is OK, I suppose. I just don't like it, and use ## explicitly.

BTW, Comeau online thingy gets it right in all modes, VC++ gets it right,
and I am betting, countless others get it right, I am just too lazy to
check.

V

[Rob Williscroft]

Re: preprocessor q: impossible macro?

Victor Bazarov wrote in news:ASSCe.1945 6$Tf5.2888
@newsread1.mlps ca01.us.to.veri o.net in comp.lang.c++:
[color=blue]
>
> Yes, you need an indirect "stringizin g" macro for that.
>
> #define STR(x) #x
> #define MYMACRO(num) "#" ## STR(num)
>[/color]

You don't need the ## above as strings concatinate anyway:

#define MYMACRO(num) ( "#" STR(num) )

Also AIUI using ## to concatinate 2 tokens that don't then
make a valid preprocessor token is illegal.
[color=blue]
> #define IDB_BUTTON 225
>
> #include <stdio.h>
>
> int main(void) {
> printf("Got %s\n", MYMACRO(IDB_BUT TON));
> }[/color]

Rob.
--

http://www.victim-prime.dsl.pipex.com/

[Victor Bazarov]

Re: preprocessor q: impossible macro?

Rob Williscroft wrote:[color=blue]
> Victor Bazarov wrote in news:ASSCe.1945 6$Tf5.2888
> @newsread1.mlps ca01.us.to.veri o.net in comp.lang.c++:
>
>[color=green]
>>Yes, you need an indirect "stringizin g" macro for that.
>>
>> #define STR(x) #x
>> #define MYMACRO(num) "#" ## STR(num)
>>[/color]
>
>
> You don't need the ## above as strings concatinate anyway:
>
> #define MYMACRO(num) ( "#" STR(num) )
>
> Also AIUI using ## to concatinate 2 tokens that don't then
> make a valid preprocessor token is illegal.[/color]

I don't understand this sentence. Perhaps you could rephrase...
Also, see my reply to Charles Mills. The code is valid.
[color=blue][color=green]
>> #define IDB_BUTTON 225
>>
>> #include <stdio.h>
>>
>> int main(void) {
>> printf("Got %s\n", MYMACRO(IDB_BUT TON));
>> }[/color]
>
>
> Rob.[/color]

V

[Me]

Re: preprocessor q: impossible macro?

Eric wrote:[color=blue]
> Hello all,
> I've got this line of given code (cannot change this; wizard-generated, but
> value may change someday):
> #define IDB_BUTTON 225
>
> Somewhere in the code I found / need this:
> ....somefunctio n("#225")....
>
> This string in the function call is the same number like in the define, but
> the constant wasn't used. Because for future changes and for readability I
> wanted to create this string with a preprocessor macro like this:
> ....somefunctio n(MYMACRO(IDB_B UTTON))....
>
> I didn't get such a macro to work. I tried:
> #define MYMACRO(num) "#" #num
> but this only yields to
> ....somefunctio n("#" "IDB_BUTTON").. ..
> I also tried with the token-pasting operator (##) etc.
>
> Is this possible at all?[/color]

Yes (see the other responses to this thread) but if you're programming
Windows like I suspect you are, you don't even need to do this, you can
just use the MAKEINTATOM(IDB _BUTTON) macro provided in windows.h which
does something more efficient based on a feature of the OS.

[Rob Williscroft]

Re: preprocessor q: impossible macro?

Victor Bazarov wrote in news:JoUCe.1946 3$Tf5.10002
@newsread1.mlps ca01.us.to.veri o.net in comp.lang.c++:
[color=blue][color=green][color=darkred]
>>> #define MYMACRO(num) "#" ## STR(num)
>>>[/color]
>>
>>
>> You don't need the ## above as strings concatinate anyway:
>>
>> #define MYMACRO(num) ( "#" STR(num) )
>>
>> Also AIUI using ## to concatinate 2 tokens that don't then
>> make a valid preprocessor token is illegal.[/color]
>
> I don't understand this sentence. Perhaps you could rephrase...[/color]

Here's n example of the use of ## pehraps it will help:

#include <iostream>
#include <ostream>

#define STR( x ) #x
#define XSTR( x ) "X" STR( x )

#define PSTR_2( x, y ) x ## STR( y )
#define PSTR( x, y ) PSTR_2(x, y )


int main()
{
std::cout << STR(i) << std::endl;
std::cout << PSTR(X,i) << std::endl;
}

The 2 tokens you are attempting to paste are " and STR, "STR
is not a valid preprocessor token.
[color=blue]
> Also, see my reply to Charles Mills. The code is valid.
>[/color]

You can find this in the Standard

16.3.3 The ## operator

...

If the result is not a valid preprocessing token,
the behavior is undefined



Rob.
--

http://www.victim-prime.dsl.pipex.com/

[Victor Bazarov]

Re: preprocessor q: impossible macro?

Rob Williscroft wrote:[color=blue]
> Victor Bazarov wrote in news:JoUCe.1946 3$Tf5.10002
> @newsread1.mlps ca01.us.to.veri o.net in comp.lang.c++:
>
>[color=green][color=darkred]
>>>> #define MYMACRO(num) "#" ## STR(num)
>>>>
>>>
>>>
>>>You don't need the ## above as strings concatinate anyway:
>>>
>>> #define MYMACRO(num) ( "#" STR(num) )
>>>
>>>Also AIUI using ## to concatinate 2 tokens that don't then
>>>make a valid preprocessor token is illegal.[/color]
>>
>>I don't understand this sentence. Perhaps you could rephrase...[/color]
>
>
> Here's n example of the use of ## pehraps it will help:
>
> #include <iostream>
> #include <ostream>
>
> #define STR( x ) #x
> #define XSTR( x ) "X" STR( x )
>
> #define PSTR_2( x, y ) x ## STR( y )
> #define PSTR( x, y ) PSTR_2(x, y )
>
>
> int main()
> {
> std::cout << STR(i) << std::endl;
> std::cout << PSTR(X,i) << std::endl;
> }
>
> The 2 tokens you are attempting to paste are " and STR, "STR
> is not a valid preprocessor token.
>
>[color=green]
>>Also, see my reply to Charles Mills. The code is valid.
>>[/color]
>
>
> You can find this in the Standard
>
> 16.3.3 The ## operator
>
> ...
>
> If the result is not a valid preprocessing token,
> the behavior is undefined[/color]

Yes, how is that applicable here? Did you read my reply to Charles?

V

[Rob Williscroft]

Re: preprocessor q: impossible macro?

Victor Bazarov wrote in news:q%UCe.1946 6$Tf5.9220
@newsread1.mlps ca01.us.to.veri o.net in comp.lang.c++:
[color=blue][color=green]
>> You can find this in the Standard
>>
>> 16.3.3 The ## operator
>>
>> ...
>>
>> If the result is not a valid preprocessing token,
>> the behavior is undefined[/color]
>
> Yes, how is that applicable here? Did you read my reply to Charles?
>
>[/color]

Yes, your argument is incorrect, reread 16.3.3, the "*all*" in
your paraphrasing ("essentialy says ...") is wrong. Only macro
paramiters are expanded before ## is used to paste tokens.

Not that that would matter anyway, the token on the left of the ##
is a quote ("), and a quote is a single token, its *never* part
of another token. Which violates the above cut&paste from the
Standard.

Rob.
--

http://www.victim-prime.dsl.pipex.com/

[Charles Mills]

Re: preprocessor q: impossible macro?

Rob Williscroft wrote:[color=blue]
> Victor Bazarov wrote in news:q%UCe.1946 6$Tf5.9220
> @newsread1.mlps ca01.us.to.veri o.net in comp.lang.c++:
>[color=green][color=darkred]
> >> You can find this in the Standard
> >>
> >> 16.3.3 The ## operator
> >>
> >> ...
> >>
> >> If the result is not a valid preprocessing token,
> >> the behavior is undefined[/color]
> >
> > Yes, how is that applicable here? Did you read my reply to Charles?
> >
> >[/color]
>
> Yes, your argument is incorrect, reread 16.3.3, the "*all*" in
> your paraphrasing ("essentialy says ...") is wrong. Only macro
> paramiters are expanded before ## is used to paste tokens.
>
> Not that that would matter anyway, the token on the left of the ##
> is a quote ("), and a quote is a single token, its *never* part
> of another token. Which violates the above cut&paste from the
> Standard.
>
> Rob.
> --
> http://www.victim-prime.dsl.pipex.com/[/color]

I think you mean a string literal is a single token, right?
So pasting two string literals like:
"a" ## "b"
would create the token:
"a""b"
which is an invalid token.
Another example would be pasting the tokens c and ++
c ## ++
creates the token
c++
which is not a valid C token :)

-Charlie

[Rob Williscroft]

Re: preprocessor q: impossible macro?

Charles Mills wrote in news:1121724926 .368427.251760
@g49g2000cwa.go oglegroups.com in comp.lang.c++:
[color=blue]
>
> I think you mean a string literal is a single token, right?[/color]

I didn't, but I was wrong :(. Thanks for the correction.
[color=blue]
> So pasting two string literals like:
> "a" ## "b"
> would create the token:
> "a""b"
> which is an invalid token.[/color]

Not so sure about that, which is maybe why Victor, believes
his interpritation is correct, if valid and *all* tokens were
expanded prior to ## doing its thing, then Victors code would
be valid. But that isn't what the Standard says, but it possibly
is how many C++ preprocessors currently work.
[color=blue]
> Another example would be pasting the tokens c and ++
> c ## ++
> creates the token
> c++
> which is not a valid C token :)
>[/color]

Agreed.

Rob.
--

http://www.victim-prime.dsl.pipex.com/

[Greg]

Re: preprocessor q: impossible macro?



Victor Bazarov wrote:[color=blue]
> Charles Mills wrote:[color=green]
> > Victor Bazarov wrote:
> >[color=darkred]
> >>Charles Mills wrote:
> >>
> >>>Victor Bazarov wrote:
> >>>[..]
> >>>
> >>>> #define STR(x) #x
> >>>> #define MYMACRO(num) "#" ## STR(num)
> >>>>
> >>>> #define IDB_BUTTON 225
> >>>>
> >>>> #include <stdio.h>
> >>>>
> >>>> int main(void) {
> >>>> printf("Got %s\n", MYMACRO(IDB_BUT TON));
> >>>> }
> >>>>
> >>>>V
> >>>
> >>>
> >>>The above solution does not compile. [..]
> >>
> >>On what compiler? What is the error message?
> >>
> >>V[/color][/color]
>
> IOW, the result of 'MYMACRO(IDB_BU TTON)' yields
>
> "#" ## STR(225)
>
> which in turn, after rescanning yields
>
> "#" ## #225
>
> which then yields
>
> "#" ## "225"
>
> And two tokens are concatenated. Relying on the catenation of strings
> is OK, I suppose. I just don't like it, and use ## explicitly.
>
> BTW, Comeau online thingy gets it right in all modes, VC++ gets it right,
> and I am betting, countless others get it right, I am just too lazy to
> check.
>
> V[/color]

Your code may compile but its output is not correct. The preprocessed
output of the printf statement in the above example is:

std::printf("Go t %s\n", "#""IDB_BUT TON" );

whereas the corrrect result would be:

std::printf("Go t %s\n", "#225" );

or its equivalent. Omitting the ## token pasting operator does yield
the correct result. Another solution would be to use slightly different
macros:

#define STR(num) "#" ## #num
#define MYMACRO(num) STR(num)

In both of these cases the preprocessed output is:

std::printf("Go t %s\n", "#" "225" );

As others have noted, the compiler will concatenate adjacent string
literals, making this output match the correct result;

Greg