Pavel Pereslegin created IGNITE-25716:
-----------------------------------------
Summary: Sql. Normalize dynamic parameter value according to
determined type
Key: IGNITE-25716
URL: https://issues.apache.org/jira/browse/IGNITE-25716
Project: Ignite
Issue Type: Bug
Components: sql, sql ai3
Reporter: Pavel Pereslegin
The following test fails
{code:sql}
LocalTime time = LocalTime.parse("00:00:00.1");
assertQuery("SELECT ?::TIME") // SELECT ?::TIME(0) produces the same result
.withParam(time)
.returns(time.withNano(0))
.check();
// Expected: 00:00 <class java.time.LocalTime>
// Actual: 00:00:00.100 <class java.time.LocalTime>
{code}
The plan is
{noformat}
Project(EXPR$0=[CAST(?0):TIME(0)]): rowcount = 1.0, cumulative cost =
IgniteCost [rowCount=2.0, cpu=2.0, memory=0.0, io=0.0, network=0.0], id = 28
Values(tuples=[[{ 0 }]]): rowcount = 1.0, cumulative cost = IgniteCost
[rowCount=1.0, cpu=1.0, memory=0.0, io=0.0, network=0.0], id = 24
{noformat}
The type of dynamic parameter is determined as TIME(0), so in this case Calcite
doesn't implement excessive cast (TIME::TIME) for this case.
A possible solution is to normalize the value of the dynamic parameter to a
determined type on initialization, so that when it is retrieved from the
execution context, it already has the expected precision.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)