[SPARK-42251][SQL] Forbid deicmal type if precision less than 1

[ulysses-you]

What changes were proposed in this pull request?

throw exception if the decimal precision less than 1.

Why are the changes needed?

Spark does not actually support decimal type with 0 precision. e.g.

– work with in-memory catalog
create table t (c decimal(0, 0)) using parquet;

-- fail with parquet
-- java.lang.IllegalArgumentException: Invalid DECIMAL precision: 0
-- at org.apache.parquet.Preconditions.checkArgument(Preconditions.java:57)
insert into table t values(0);

-- fail with hive catalog
-- Caused by: java.lang.IllegalArgumentException: Decimal precision out of allowed range [1,38]
-- at org.apache.hadoop.hive.serde2.typeinfo.HiveDecimalUtils.validateParameter(HiveDecimalUtils.java:44)
create table t (c decimal(0, 0)) using parquet;

Does this PR introduce any user-facing change?

yes, one main behavior change is: SELECT cast(0 as decimal(0, 0))

• before: return 0
• after: throw exception

How was this patch tested?

add test

[ulysses-you]

I'm not sure this worth a leagcy config. cc @cloud-fan @viirya @gengliangwang

[cloud-fan]

how does cast(0 as decimal(0, 0)) work?

[cloud-fan]

Another place to check is user-specified schema when reading data sources, e.g. spark.read.schema("c1: decimal(0, 0)").json(...)

[ulysses-you]

how does cast(0 as decimal(0, 0)) work?

It's wrapped in Decimal first which internally use longVal to represent the decimal value so it won't fail. The changePrecision method only fail with decimal(0, 0) if it uses decimalVal(BigDecimal). That's the triky place that it may hide bug since some operators do not fail.

Another place to check is user-specified schema when reading data source

This case should be covered. It's similar with create table ... ddl. I added a new test for it.

[github-actions]

We're closing this PR because it hasn't been updated in a while. This isn't a judgement on the merit of the PR in any way. It's just a way of keeping the PR queue manageable.
If you'd like to revive this PR, please reopen it and ask a committer to remove the Stale tag!