CLK_TCK undefined

CLK_TCK undefined

[Andrea Monaco]

The constant CLK_TCK seems not defined in my glibc headers, although
it's required by C89 (according to the draft I'm using).  Is that
intended?



Andrea Monaco

Re: CLK_TCK undefined

[Florian Weimer]

* Andrea Monaco via Glibc-bugs:

> The constant CLK_TCK seems not defined in my glibc headers, although
> it's required by C89 (according to the draft I'm using).  Is that
> intended?

We have a conditional definition in <time.h>:

#if (!defined __STRICT_ANSI__ || defined __USE_POSIX) \
   && !defined __USE_XOPEN2K
/* Even though CLOCKS_PER_SEC has such a strange value CLK_TCK
   presents the real value for clock ticks per second for the system.  */
extern long int __sysconf (int);
# define CLK_TCK ((__clock_t) __sysconf (2))    /* 2 is _SC_CLK_TCK */
#endif

Maybe your choice of feature selection macros suppresses the definition?

Thanks,
Florian

Re: CLK_TCK undefined

[Andreas Schwab]

On Okt 16 2023, Andrea Monaco via Glibc-bugs wrote:

> it's required by C89 (according to the draft I'm using).

Where does it say that?

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510  2552 DF73 E780 A9DA AEC1
"And now for something completely different."

Re: CLK_TCK undefined

[Adhemerval Zanella Netto]

On 18/10/23 06:55, Florian Weimer via Glibc-bugs wrote:
> * Andrea Monaco via Glibc-bugs:
> 
>> The constant CLK_TCK seems not defined in my glibc headers, although
>> it's required by C89 (according to the draft I'm using).  Is that
>> intended?
> 
> We have a conditional definition in <time.h>:
> 
> #if (!defined __STRICT_ANSI__ || defined __USE_POSIX) \
>    && !defined __USE_XOPEN2K
> /* Even though CLOCKS_PER_SEC has such a strange value CLK_TCK
>    presents the real value for clock ticks per second for the system.  */
> extern long int __sysconf (int);
> # define CLK_TCK ((__clock_t) __sysconf (2))    /* 2 is _SC_CLK_TCK */
> #endif
> 
> Maybe your choice of feature selection macros suppresses the definition?

It was removed by POSIX 2001 [1]:

  Issue 6

  The symbolic name CLK_TCK is removed. Application usage is added describing 
  how its equivalent functionality can be obtained using sysconf().

You need to specify an old POSIX version to actually use it:

$ cat t.c
#include <time.h>
#include <stdio.h>
#include <stdint.h>

int main (int argc, char *argv[])
{
  printf ("%jd\n", (intmax_t) CLK_TCK);
  return 0;
}

$ gcc -Wall t.c -o t -D_POSIX_C_SOURCE=200112L
t.c: In function ‘main’:
t.c:7:31: error: ‘CLK_TCK’ undeclared (first use in this function)
    7 |   printf ("%jd\n", (intmax_t) CLK_TCK);
      |                               ^~~~~~~
t.c:7:31: note: each undeclared identifier is reported only once for each function it appears in
$ gcc -Wall t.c -o t -D_POSIX_C_SOURCE=199506
$ ./t
100


[1] https://pubs.opengroup.org/onlinepubs/9699919799/basedefs/time.h.html

Re: CLK_TCK undefined

[Andrea Monaco]

  > It was removed by POSIX 2001 [1]:

I understand.  But still, it was just a harmless constant, so removing
it is quite strong in my opinion, since glibc usually doesn't just
follow POSIX blindly.


  > Maybe your choice of feature selection macros suppresses the
  > definition?

I don't define any feature macros, but I pass the -ansi option.  Also, I
don't know if autoconf adds other definitions.


  > Where does it say that?

http://jfxpt.com/library/c89-draft.html, search for CLK_TCK.



Andrea Monaco

Re: CLK_TCK undefined

[Andreas Schwab]

On Okt 19 2023, Andrea Monaco wrote:

> http://jfxpt.com/library/c89-draft.html, search for CLK_TCK.

In the final standard it has been replaced by CLOCKS_PER_SEC.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510  2552 DF73 E780 A9DA AEC1
"And now for something completely different."

Re: CLK_TCK undefined

[Andrea Monaco]

  > In the final standard it has been replaced by CLOCKS_PER_SEC.

Really?  Have you got a source?



Andrea Monaco

Re: CLK_TCK undefined

[Joseph Myers]

On Thu, 19 Oct 2023, Andrea Monaco via Glibc-bugs wrote:

>   > In the final standard it has been replaced by CLOCKS_PER_SEC.
> 
> Really?  Have you got a source?

https://nvlpubs.nist.gov/nistpubs/Legacy/FIPS/fipspub160.pdf#page=185

-- 
Joseph S. Myers
joseph@codesourcery.com