change date_part return types to f64

[waitingkuo]

date_part currently returns i32 which might lose some information, e.g.

❯ select date_part('second', timestamp '2000-01-01T00:00:00.1');
+--------------------------------------------------------+
| datepart(Utf8("second"),Utf8("2000-01-01T00:00:00.1")) |
+--------------------------------------------------------+
| 0                                                      |
+--------------------------------------------------------+
1 row in set. Query took 0.000 seconds.

while postgresql returns double precision

willy=# select date_part('second', timestamp '2000-01-01T00:00:00.1');
 date_part 
-----------
       0.1
(1 row)

is it recommended to follow postgresql's return type?
if it is preferred, perhaps we could do a separate pr to deal with this

change it to double precision

Describe the solution you'd like
A clear and concise description of what you want to happen.

Describe alternatives you've considered
A clear and concise description of any alternative solutions or features you've considered.

perhaps we could consider return decimals instead to align with extract to return decimal. postgres doc mentions this:

For historical reasons, the date_part function returns values of type double precision. 
This can result in a loss of precision in certain uses. Using extract is recommended instead.

https://www.postgresql.org/docs/current/functions-datetime.html

Additional context
Add any other context or screenshots about the feature request here.

[comphead]

@waitingkuo may I try this?

[waitingkuo]

@waitingkuo may I try this?

sure, please try it~

[comphead]

@waitingkuo the problem is date_part calls for arrow::compute::kernels::temporal::second which is i32, https://docs.rs/arrow/latest/arrow/compute/kernels/temporal/fn.second.html and they dont have millisecond support. Do you think its good idea to implement milliseconds by ourselves?

[waitingkuo]

@comphead

arrow-rs uses chrono-rs to extract these date parts. Chono-rs does have nanosecond while arrow-rs doesn't have.

We could extend arrow-rs to support nanosecond and then get back to this issue.

I think adding nanosecond support in arrow-rs is a good start issue. If you're interested in, I'm happy to help you to figure out the codes to modify. Basically, you could follow the second's logic to do so
https://github.com/apache/arrow-rs/blob/master/arrow/src/compute/kernels/temporal.rs#L723-L764

I could help to send a pr to arrow-rs as well

[waitingkuo]

cc @alamb

[comphead]

@waitingkuo thanks for the details, let me try to add the ticket and PR to arrow-rs.

[comphead]

Related arrow-rs ticket apache/arrow-rs#2995

[comphead]

@waitingkuo the apache/arrow-rs#2995 merged, should we wait for next arrow-rs version? How it usually happens?

[waitingkuo]

@comphead thank you. i think we need to wait for 27.0.0 . 26.0.0 is just fixed. It normally released 1 or 2 times a month

[comphead]

waiting for #4199