Implement `current_time` Function

[alamb]

Is your feature request related to a problem or challenge? Please describe what you are trying to do.
As we add more full featured date/time support for DataFusion users need to be able to use the current time to calculate things like "all data in the last 5 hours"

Describe the solution you'd like
Support the current_time postgres function https://www.postgresql.org/docs/current/functions-datetime.html#FUNCTIONS-DATETIME-CURRENT

Note I think we should avoid the current_time(precision) variant initially -- we can implement it as a follow on PR.

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

Additional context
This is a feature request we got from an early (internal) user of InfluxDB IOx

I believe this is a good first issue as you can follow the existing implementation of now()

• Physical support: https://github.com/apache/arrow-datafusion/blob/f1322592046a7ea895820286a2ad16d934540d92/datafusion/physical-expr/src/functions.rs#L424-L429) and use the same query_execution_start_time field
• Tests: https://github.com/apache/arrow-datafusion/blob/beeb631d0bb979e9b6826f6cd3eae2d0a24f471e/datafusion/core/tests/sql/timestamp.rs#L417

It would be a good exercise to add a new build in scalar function, physical planing support, and then tests

Note this the same ticket exists for the current_date function #3981

[naosense]

This sounds good start to me, could you please assign it to me or some extra advises?

[waitingkuo]

@alamb postgresql's current_time returns TIMETZ

willy=# select pg_typeof(current_time);
      pg_typeof      
---------------------
 time with time zone
(1 row)

what should we return here?

[alamb]

@waitingkuo I was imagining it would return a https://docs.rs/arrow/latest/arrow/array/type.Time64NanosecondArray.html

But I have not thought as deeply as you about the timezone implications, so if you have a different suggestion I would love to hear it.

This sounds good start to me, could you please assign it to me or some extra advises?

Hi @naosense -- thanks and done!

[comphead]

This sounds good start to me, could you please assign it to me or some extra advises?

Hi @naosense please follow the #4022 PR

[waitingkuo]

@alamb i have no idea for timetz in arrow-rs for now as Time has not timezone (e.g. DataType::Time64(TimeUnit::Nanosecond) while Timestamp has (e.g. DataType::Timestamp(TimeUnit::Nanosecond, Some("UTC".to_string()))

i think we could either

1. follow postgresql's function signature that means we could implement LOCAL_TIME instead
2. keep it simple, return TIME directly for now and comment it

in the long run, we might consider make these function names explicitly like what was discussed in
#686 (comment)

[naosense]

This sounds good start to me, could you please assign it to me or some extra advises?

Hi @naosense please follow the #4022 PR

That's nice, thanks!

[alamb]

keep it simple, return TIME directly for now and comment it

This would be my suggestion