Tracking: Support DataType::Decimal

[sundy-li]

Summary

Description for this feature.

Now we have introduced Decimal128 and Decimal256 types, let's add more features on decimal types.

• feat(parser): parse decimal type #9894
• feat(expr): add Decimal128 and Decimal256 type #9856
• feat(decimal): add operation for decimal #9926 supports number -> Decimal, Decimal -> Decimal and arithmetic functions @sundy-li
• feat(query): add compare op for decimal types #10214
• Group KeysU128 & KeysU256 will be stored as Decimal @sundy-li
• feat(query): add sum/avg for decimal types #10059
• Add min/max aggregate function on Decimal, support order by Decimal datatypes @TCeason #10005
• Add Decimal column stats to segment info (feat(query): support desc table with decimal type and support order by decmail type col #10113)
• Decimal literal/encoding @youngsofun
  • Insert values feat: support insert decimal. #10167
    • may fallback to expr of parser
  • Insert/copy with other formats feat: load/unload decimal with CSV/TSV/NDJSON #10176
  • Cast from string or. allow string directly feat: support function to_decimal(string). #10369
  • Used decimal for number literal by default instead of float Feature: numeric constants literal should be parsed as fixed-point numbers instead of floating-point numbers to avoid losing precision without notice  #10220
  • deal with the float in json. Float -> string -> decimal. or parse json by ourself
  • binary format used inside.
• feat(doc): add decimal docs #10095

Below is old comments:

Decimal is Synonymous with Number type.

It represents a range up to 38 digits, with an optional precision and scale:

Precision:
Total number of digits allowed.

Scale:
Number of digits allowed to the right of the decimal point.

By default Number means NUMBER(38,0) or Decimal(38, 0)

Width	Internal	Size (Bytes)
1-4	INT16	2
5-9	INT32	4
10-18	INT64	8
19-38	INT128	16

[liukun4515]

Hi @sundy-li, I have a question about decimal data type which will be implemented in the feature.
From the datatype, we know the native data type(i32,i64) is simple. when add the decimal, we need to consider the precision and scale, refine the data type coercion.

Does the community has the plan to implement it？

[sundy-li]

How about making decimal's physical_type to Int128 ?
As arrow did https://github.com/jorgecarleitao/arrow2/blob/main/src/datatypes/mod.rs#L247

[youngsofun]

do we need to/ able to support Decimal when check results column types in the sql logical test?

[sundy-li]

do we need to/ able to support Decimal when check results column types in the sql logical test?

Decimal will be formatted into string, so it's same as check string result.

[BohuTANG]

@sundy-li

We also need a task: put the decimal-type column stats to segment to fast pruning.