Clarification on future direction of POSIXRULES and POSIX TZ specs?
The 2019b announcement says that zic's -p feature is being deprecated with an eye to eventual removal. I am wondering what is the game plan there exactly, and in particular what is planned to happen in localtime.c. Will the code that tries to load the TZDEFRULES zone be removed? If so, will the behavior for transition-rule-less POSIX TZ specs be that it always falls back to the hard-wired TZDEFRULESTRING? Or are you planning to go further, say by rejecting such TZ specs entirely? (As far as I can see, doing that would be contrary to the POSIX spec --- although they don't say what the default rule should be, it's pretty clear syntactically that you can omit it.) Also, what sort of time scale might this happen on? regards, tom lane
Tom Lane wrote:
will the behavior for transition-rule-less POSIX TZ specs be that it always falls back to the hard-wired TZDEFRULESTRING?
That sounds like a good way forward, yes. It conforms to POSIX, it corresponds to the intent of the ancient Unix System V code, and for timestamps after 2007 it matches the current tzdb default behavior.
participants (2)
-
Paul Eggert -
Tom Lane