Re: [PATCH v2 7/7] misc: syslog: Use RFC5424

[Adhemerval Zanella <adhemerval.zanella@linaro.org>]

On 19/02/2022 00:50, Paul Eggert wrote:
> There should be documentation about the change, somewhere.
> 
> On 2/18/22 06:23, Adhemerval Zanella via Libc-alpha wrote:
> 
>> +  struct tm *gmtm = __gmtime64_r (&ts.tv_sec, &now_tm);
>> +  if (gmtm == NULL)
>> +    strcpy (timebuf->b, NILVALUE);
> 
> RFC 5424 requires unsigned four-digit years, so change "gmtm == NULL" to "gmtm == NULL || ! (-1900 <= gmtm->tm_year && gmtm->tm_year <= 9999 - 1900)". Or, if the idea is to depart from RFC 5424 for out-of-range years (which might be better), document the departure.

Indeed I don't see why we can't support out-of-range years in such case.  I will add
a comment.

> 
>> +  else
>> +    {
>> +      size_t datebytes = __strftime_l (timebuf->b, sizeof timebuf->b,
>> +                       "%FT%T", gmtm, _nl_C_locobj_ptr);
> RFC 5424 does not allow leap seconds, so something like this:
> 
>     if (gmtm->tv_sec == 60)
>       {
>          gmtm->tv_sec = 59;
>          gmtm->tv_nsec = 999999999;
>       }
> 
> before calling __strftime_l. (Or again, if the idea is to depart from RFC 5424, document the departure.)

I think this is worth to support, I will add such test.

> 
>> +      int usec = ts.tv_nsec / 1000;
> 
> Typically fewer instructions (and still correct) if you replace 1000 with 1000ul (admittedly a microoptimization....).

Ack.