refactor(expr): decouple macro gen_binary_expr_atm from temporal operator signatures

[xiangjinwu]

What's changed and what's your intention?

The macro gen_binary_expr_atm currently list certain combinations of temporal types, e.g. timestamp_timestamp_f, timestamp_interval_f. This makes it harder to add more temporal operators as their signatures are not that regular compared to numbers. It makes less sense to reuse them across add/sub/mul/div. For details, check the pg operator table below.

This PR allows gen_binary_expr_atm to pass thru a list to gen_atm_impl directly, so that different operators (add/sub/mul/div) can add their own implementations without touching the definition of gen_binary_expr_atm any more.

     lhs     | oprname |     rhs     | ?column? |     ret     |         oprcode         
-------------+---------+-------------+----------+-------------+-------------------------
 float8      | *       | interval    | =        | interval    | mul_d_interval
 interval    | *       | float8      | =        | interval    | interval_mul
 date        | +       | int4        | =        | date        | date_pli
 int4        | +       | date        | =        | date        | integer_pl_date
 interval    | +       | interval    | =        | interval    | interval_pl
 interval    | +       | time        | =        | time        | interval_pl_time
 time        | +       | interval    | =        | time        | time_pl_interval
 date        | +       | interval    | =        | timestamp   | date_pl_interval
 date        | +       | time        | =        | timestamp   | datetime_pl
 interval    | +       | date        | =        | timestamp   | interval_pl_date
 interval    | +       | timestamp   | =        | timestamp   | interval_pl_timestamp
 time        | +       | date        | =        | timestamp   | timedate_pl
 timestamp   | +       | interval    | =        | timestamp   | timestamp_pl_interval
 interval    | +       | timestamptz | =        | timestamptz | interval_pl_timestamptz
 timestamptz | +       | interval    | =        | timestamptz | timestamptz_pl_interval
 date        | -       | int4        | =        | date        | date_mii
 date        | -       | date        | =        | int4        | date_mi
 interval    | -       | interval    | =        | interval    | interval_mi
 time        | -       | time        | =        | interval    | time_mi_time
 timestamp   | -       | timestamp   | =        | interval    | timestamp_mi
 timestamptz | -       | timestamptz | =        | interval    | timestamptz_mi
 time        | -       | interval    | =        | time        | time_mi_interval
 date        | -       | interval    | =        | timestamp   | date_mi_interval
 timestamp   | -       | interval    | =        | timestamp   | timestamp_mi_interval
 timestamptz | -       | interval    | =        | timestamptz | timestamptz_mi_interval
 interval    | /       | float8      | =        | interval    | interval_div
(26 rows)

Query:

with min_types as (
    select * from pg_type
    where typname in (
        'bool', 'int2', 'int4', 'int8', 'float4', 'float8', 'numeric',
        'text', 'date', 'time', 'timestamp', 'timestamptz', 'interval'
    )
) select
    type_l.typname as lhs,
    oprname,
    type_r.typname as rhs,
    '=',
    type_o.typname as ret,
    oprcode
from pg_operator
    join min_types as type_l on pg_operator.oprleft = type_l.oid
    join min_types as type_r on pg_operator.oprright = type_r.oid
    join min_types as type_o on pg_operator.oprresult = type_o.oid
where oprname in ('+', '-', '*', '/')
and (
    type_l.typname in ('date', 'timestamp', 'timestamptz', 'time', 'interval')
    or type_r.typname in ('date', 'timestamp', 'timestamptz', 'time', 'interval')
)
order by oprname, ret, lhs, rhs

Checklist

• I have written necessary docs and comments
  - [ ] I have added necessary unit tests and integration tests (Refactor does not break existing tests.)

Refer to a related PR or issue link (optional)

[codecov]

Codecov Report

Merging #1985 (ca67afe) into main (06e0f62) will decrease coverage by 0.01%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##             main    #1985      +/-   ##
==========================================
- Coverage   70.95%   70.93%   -0.02%     
==========================================
  Files         625      625              
  Lines       80587    80656      +69     
==========================================
+ Hits        57177    57211      +34     
- Misses      23410    23445      +35     

Flag	Coverage Δ
rust	70.93% <ø> (-0.02%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
src/expr/src/expr/expr_binary_nonnull.rs	83.57% <ø> (ø)
src/meta/src/stream/stream_manager.rs	69.82% <0.00%> (-2.68%)	⬇️
.../stream/src/executor/managed_state/flush_status.rs	93.87% <0.00%> (-1.96%)	⬇️
src/meta/src/stream/fragmenter.rs	84.26% <0.00%> (-1.64%)	⬇️
src/meta/src/stream/graph/stream_graph.rs	60.50% <0.00%> (-1.44%)	⬇️
src/connector/src/filesystem/file_common.rs	80.44% <0.00%> (-0.45%)	⬇️
...ecutor/managed_state/top_n/top_n_bottom_n_state.rs	79.25% <0.00%> (-0.28%)	⬇️
...am/src/executor/managed_state/top_n/top_n_state.rs	92.65% <0.00%> (-0.11%)	⬇️
src/meta/src/stream/test_fragmenter.rs	99.62% <0.00%> (ø)
src/meta/src/rpc/service/ddl_service.rs	0.00% <0.00%> (ø)
... and 7 more

📣 Codecov can now indicate which changes are the most critical in Pull Requests. Learn more

[skyzh]

LGTM