Re: [HACKERS] Bug in to_timestamp().

On Sun, Jul 22, 2018 at 08:22:23AM -0700, David G. Johnston wrote:
> My re-read of the thread the other day left me with a feeling of
> contentment that this was an acceptable change but I also get the feeling
> like I'm missing the downside trade-off too...I was hoping your review
> would help in that regard but as it did not speak to specific
> incompatibilities it has not.

I like more behaviour of the function with the patch. It gives less
unexpected results. For example, the query mentioned above:

SELECT to_timestamp('2011-12-18 23:38:15', 'YYYY-MM-DD  HH24:MI:SS')

I looked for some tradeoffs of the patch. I think it could be parsing
strings like the following input strings:

SELECT TO_TIMESTAMP('2011年5月1日', 'yyyy-MM-DD');
SELECT TO_TIMESTAMP('2011y5m1d', 'yyyy-MM-DD');

HEAD extracts year, month and day from the string. But patched
to_timestamp() raises an error. Someone could rely on such behaviour.
The patch divides separator characters from letters and digits. And
'年' or 'y' are letters here. And so the format string doesn't match the
input string.

-- 
Arthur Zakirov
Postgres Professional: http://www.postgrespro.com
Russian Postgres Company