>No concern that unexpected behavior might result from processes picking up
>dynamic changes to timezone rules?  If not, how about caching the last time_t 
>passed to
>localtime() and only checking if it had changed (or changed by some amount)?
>(although that could get ugly for localtime_r(), which I suppose would need
>read/write locks on the static variable)


Well, if the $TZ is set to something like "US/Pacific" and the
politicians feel the need to "act on the evironment" again, then I
would expect that such a timezone file update only leads to
different outcomes for future times.  And so no inconsistencies
will be seen.

(In principle, one can argue that it is wrong to use any form
of timezone calculations for times in the future, because of
politicians)


But if $TZ is set to "localtime" and the system administrator
chnages the system's timezone, then that may be a different matter;
again, only cron would get really upset, I think.

>At least until efficient userland file event monitoring facilities are 
>available...

Indeed.

Casper
_______________________________________________
opensolaris-discuss mailing list
[email protected]

Reply via email to