Josh Berkus <josh@agliodbs.com> writes:
> And, on the up side, if the various *nixes fix their time zone behaviour past
> 2037, then Postgres will be automatically fixed as well, yes?
Depends what the new API looks like. I would think that a sane answer
is to redefine time_t as a signed 64-bit value, preserving the 1/1/1970
zero origin, but who knows what the library people will really do?
glibc's recent move to redefine time_t as unsigned (losing support for
all pre-1970 dates) doesn't give me high confidence in their design
sensibility.
In any case you can bet that we'll have some work to do when a better
API is available.
regards, tom lane