Re:Re: Re: Re: Re:Re: [BUGS] Re: [BUGS] Return value error of‘to_timestamp’

在 2016-08-18 22:09:06，"Tom Lane" <tgl@sss.pgh.pa.us> 写道：
>Francisco Olarte <folarte@peoplecall.com> writes:
>> On Thu, Aug 18, 2016 at 11:57 AM, 甄明洋 <zhenmingyang@yeah.net> wrote:
>>> Why don't use a unified time zone Convention ?
>
>> Given Tom said:
>>>> Aren't standards fun?
>
>> I suspect SQL std. mandates it.
>
>The SQL standard does mandate use of ISO convention for timestamp values.
>However, the use of any sort of timezone name in "SET timezone" is outside
>the SQL standard (or at least it was last I looked).  Our timezone name
>support is based on the IANA (nee Olson) timezone data set, which is used
>by just about everybody except Microsoft, and that follows the POSIX
>standard.
>
>In principle we could hack up the IANA code and data so that zone names
>that look like POSIX names follow the ISO sign convention, but if you
>ask me that's just nuts.  It would mean for example that "set timezone
>to 'PST8PDT'" inside PG would act completely differently from "TZ=PST8PDT"
>in the shell.  That would result in more confusion not less.
>
>In short, neither of these choices were made in a vacuum.
>
>			regards, tom lane