Ruben Q L created CALCITE-5977:
----------------------------------
Summary: RexLiteral returns the same value for TIMESTAMP and
TIMESTAMP_WITH_LOCAL_TIME_ZONE
Key: CALCITE-5977
URL: https://issues.apache.org/jira/browse/CALCITE-5977
Project: Calcite
Issue Type: Bug
Components: core
Reporter: Ruben Q L
Perhaps I'm missing something, but this seems odd to me.
If we want to get the value in Long format (i.e. milliseconds since 1970-01-01
00:00:00) of a TIMESTAMP / TIMESTAMP_WITH_LOCAL_TIME_ZONE RexLiteral, their
code is exactly the same (even if they are in different "case" blocks):
{code}
public @Nullable Object getValue2() {
...
switch (typeName) {
...
case TIMESTAMP:
case TIMESTAMP_WITH_LOCAL_TIME_ZONE:
return getValueAs(Long.class);
...
}
public <T> @Nullable T getValueAs(Class<T> clazz) {
...
switch (typeName) {
...
case TIMESTAMP:
if (clazz == Long.class) {
// Milliseconds since 1970-01-01 00:00:00
return clazz.cast(((TimestampString) value).getMillisSinceEpoch());
}
...
break;
case TIMESTAMP_WITH_LOCAL_TIME_ZONE:
if (clazz == Long.class) {
// Milliseconds since 1970-01-01 00:00:00
return clazz.cast(((TimestampString) value).getMillisSinceEpoch());
}
...
{code}
In case of a TIMESTAMP_WITH_LOCAL_TIME_ZONE, shouldn't this code include some
extra processing to "shift" the value with the proper offset / daylight savings?
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
