[PATCH v5] time: Add TIME_MONOTONIC, TIME_ACTIVE, and TIME_THREAD_ACTIVE

[Adhemerval Zanella Netto]

On 02/12/25 15:26, Paul Eggert wrote:
> On 2025-12-02 04:24, Adhemerval Zanella wrote:
> 
>> +  _Static_assert (TIME_UTC - 1 == CLOCK_REALTIME,
>> +          "TIME_UTC -1 == CLOCK_REALTIME");
> 
> Missing space after "-" in the string. There is another instances of this.
> 
> Better yet, include <verify.h> and just say "verify (TIME_UTC - 1 == CLOCK_REALTIME);". Two-argument _Static_assert is a useless typo magnet.

Ack, I forgot that we have this macro.

> 
> 
>> +The base @code{TIME_MONOTONIC} returns a monotonic-increasing time since
> 
> It's not necessarily increasing. Also adverb-adjective is more common in English than adjective-adjective here. So please say "monotonically nondecreasing time".
> 
> 
>> +an unspecified point in the past.
> 
> Add " that may change if the system is rebooted or suspended" (same warning as for CLOCK_MONOTONIC).

Ack.

> 
> 
>> -  if (ret == TIME_UTC)
>> +  int ret = __timespec_get64 (&tp64, base);
>> +  if (ret != 0 && ts != NULL)
> 
> Why is the "&& ts != NULL" added?  (There are two instances of this.)
> 
> The C standard doesn't require these two additions: although timespec_getres has well-defined behavior when TS is null, timespec_get does not. Let's stick to the standard rather than adding a bit of bloat to support an extension that nobody will (or should) use.
> 

Ack.