Map MySQL TIMESTAMP types to TIMESTAMP WITH TIME ZONE #18470

adamjshook · 2023-07-31T19:27:18Z

Description

This PR changes the MySQL type mappings for MySQL's TIMESTAMP type to Trino's TIMESTAMP WITH TIME ZONE type. Note that MySQL DATETIME types are read as TIMESTAMP types and Trino TIMESTAMP types are inserted into MySQL as DATETIME types -- this behavior is unchanged.

In other words, for reads and writes:

MySQL Type	Trino Type
DATETIME	TIMESTAMP
TIMESTAMP	TIMESTAMP WITH TIME ZONE

Additional context and related issues

Fixes #15668

Changed read and write mappings for MySQL TIMESTAMP type to map to Trino TIMESTAMP WITH TIME ZONE
- The implementations were taken from the PostgreSQL connector -- I am open to any suggestions on implementation details here
Added override of getColumnDefinitionSql to explicitly declare a column as NULL to satisfy an Invalid default value failure in the data mapping smoke tests when creating a table using a MySQL TIMESTAMP type
Added a separate test class for the time/date/timestamp/timestamp with time zone types where the MySQL container's time zone is set to Pacific/Apia (and therefore the time zone of the MySQL server)

Release notes

( ) This is not user-visible or docs only and no release notes are required.
( ) Release notes are required, please propose a release note for me.
(X) Release notes are required, with the following suggested text:

# MySQL Connector
* Changes MySQL TIMESTAMP types to map to TIMESTAMP WITH TIME ZONE

plugin/trino-mysql/src/main/java/io/trino/plugin/mysql/MySqlClient.java

hashhar · 2023-08-02T05:26:34Z

plugin/trino-mysql/src/main/java/io/trino/plugin/mysql/MySqlClient.java

+        return new LongWriteFunction()
+        {


Yeah, formatter did its thing and I missed it on a rebase. I'll revert.

plugin/trino-mysql/src/test/java/io/trino/plugin/mysql/BaseMySqlConnectorTest.java

hashhar · 2023-08-02T05:30:46Z

...trino-mysql/src/test/java/io/trino/plugin/mysql/TestMySqlTimeMappingsWithServerTimeZone.java

+/**
+ * Default MySQL time zone is set to UTC. This is to test the date and time type mappings when the server has a different time zone.
+ */
+public class TestMySqlTimeMappingsWithServerTimeZone


I don't see a way to combine this with existing test class anyway since the server zone needs to be changed and if done on the fly it might lead to test isolation issues.

@wendigo @dominikzalewski look at this specifically - I couldn't think of solutions but maybe there are.

@hashhar, is this a problem? Usually in other projects I was participating in, we tended to keep line numbers on 3 digits. It wasn't an error to have more than one test class for a feature.

The concern is that if you add additional cases to the other class how do you ensure it gets tested in this class as well?

A simple solution is adding a comment on the original class.

I'm just wondering if someone has other ideas. If no ideas then what we have right now is fine.

hashhar · 2023-08-02T05:31:28Z

plugin/trino-mysql/src/test/java/io/trino/plugin/mysql/TestMySqlTypeMapping.java

+                // after epoch (MySQL's timestamp type doesn't support values <= epoch)
+                .addRoundTrip("timestamp(3) WITH TIME ZONE", "TIMESTAMP '2019-03-18 10:01:17.987 %s'".formatted(sessionZone), createTimestampWithTimeZoneType(3), "TIMESTAMP '2019-03-18 10:01:17.987 UTC'")


add a separate test method verifying that the failure happens as expected (e..g it's a good user facing error).

hashhar · 2023-08-02T05:32:14Z

plugin/trino-mysql/src/test/java/io/trino/plugin/mysql/TestMySqlTypeMapping.java

+                .addRoundTrip("timestamp(1) WITH TIME ZONE", "TIMESTAMP '2020-09-27 12:34:56.9 %s'".formatted(sessionZone), createTimestampWithTimeZoneType(1), "TIMESTAMP '2020-09-27 12:34:56.9 UTC'")
+                .addRoundTrip("timestamp(3) WITH TIME ZONE", "TIMESTAMP '2020-09-27 12:34:56.123 %s'".formatted(sessionZone), createTimestampWithTimeZoneType(3), "TIMESTAMP '2020-09-27 12:34:56.123 UTC'")
+                .addRoundTrip("timestamp(3) WITH TIME ZONE", "TIMESTAMP '2020-09-27 12:34:56.999 %s'".formatted(sessionZone), createTimestampWithTimeZoneType(3), "TIMESTAMP '2020-09-27 12:34:56.999 UTC'")
+                .addRoundTrip("timestamp(6) WITH TIME ZONE", "TIMESTAMP '2020-09-27 12:34:56.123456 %s'".formatted(sessionZone), createTimestampWithTimeZoneType(6), "TIMESTAMP '2020-09-27 12:34:56.123456 UTC'")


also test values where rounding needs to be performed - important to test since write path is different for CTAS and INSERT.

Added testTimestampWithTimeZoneCoercion, a copy of testTimestampCoercion but slightly modified to to use values that are within range of MySQL TIMESTAMP

adamjshook · 2023-08-03T16:49:04Z

I added a test case to check the unsupported timestamptz values. It passes just fine when the server time zone is set to UTC, but when I change it to Pacific/Apia I don't get an error inserting 1970-01-01 00:00:00 UTC into MySQL, which makes me think something is wrong (and is why the CI is failing). Looking in that.

adamjshook · 2023-08-23T16:42:24Z

Looking into the failing test case more.

I ran the test case with the MySQL TZ set to Pacific/Apia and set some breakpoints.

Trino executes:

INSERT INTO tpch.test_unsupported_timestamptr94g5tht8 VALUES (TIMESTAMP '1970-01-01 00:00:00 UTC')

When sent to the JDBC client, the new Timestamp value is set to 0. This query succeeds when I expect it to fail.

On the server side, running the mysql CLI inside the Docker container:

bash-4.4# echo $TZ     
Pacific/Apia

mysql> SELECT @@global.time_zone, @@session.time_zone;
+--------------------+---------------------+
| @@global.time_zone | @@session.time_zone |
+--------------------+---------------------+
| SYSTEM             | SYSTEM              |
+--------------------+---------------------+
mysql> select * from test_unsupported_timestamptr94g5tht8;
+---------------------+
| data                |
+---------------------+
| 1970-01-01 00:00:00 |
+---------------------+
ysql> insert into test_unsupported_timestamptr94g5tht8 values ('1970-01-01 00:00:00');
Query OK, 1 row affected (0.03 sec)
mysql> insert into test_unsupported_timestamptr94g5tht8 values ('1969-12-31 13:00:00');
ERROR 1292 (22007): Incorrect datetime value: '1969-12-31 13:00:00' for column 'data' at row 1
mysql> insert into test_unsupported_timestamptr94g5tht8 values ('1969-12-31 13:00:01');
Query OK, 1 row affected (0.03 sec)

MySQL seems to be acting appropriately and rejecting the values after converting the given values from the session time zone to UTC. This seems like possibly a JDBC driver issue (or how we are using the JDBC driver).

hashhar · 2023-08-23T16:51:50Z

well, let's add validation like the ClickHouse connector does within the connector code itself. We can follow-up to see if we're using the driver wrong or it's just a driver bug.

hashhar · 2023-08-24T04:53:49Z

FYI @findepi @anusudarsan since you were interested in this.

dominikzalewski · 2023-08-24T11:29:56Z

plugin/trino-mysql/src/main/java/io/trino/plugin/mysql/TrinoToMySqlWriteChecker.java

+public class TrinoToMySqlWriteChecker<T>
+{
+    // The range for TIMESTAMP values is '1970-01-01 00:00:01.000000' to '2038-01-19 03:14:07.499999'
+    public static final TrinoToMySqlWriteChecker<Instant> TIMESTAMP = new TrinoToMySqlWriteChecker<>(


Is the name 'TIMESTAMP' something that has been overlooked, or a thought through variable name?

We also seem to be using guice modules. Wouldn't it be cleaner if we configured a binding 'toInstance' in guice and injected it, rather than rely on statically initialized fields?

The name is thought through - see the point of usage - TIMESTAMP.validate(value).

Note that this class is "derived" from https://github.com/trinodb/trino/blob/5d8824edf9101eab451128bbad53d8c6ce930ad4/plugin/trino-clickhouse/src/main/java/io/trino/plugin/clickhouse/TrinoToClickHouseWriteChecker.java.

And this class should go away if/when the MySQL JDBC driver bug gets fixed.

hashhar

@adamjshook Does mysql.md (the docs) need to be updated as well to mention the new mapping?

dominikzalewski · 2023-08-24T11:44:13Z

plugin/trino-mysql/src/main/java/io/trino/plugin/mysql/TrinoToMySqlWriteChecker.java

+{
+    // The range for TIMESTAMP values is '1970-01-01 00:00:01.000000' to '2038-01-19 03:14:07.499999'
+    public static final TrinoToMySqlWriteChecker<Instant> TIMESTAMP = new TrinoToMySqlWriteChecker<>(
+            new TimestampWriteValueChecker(new Range<>(Instant.parse("1970-01-01T00:00:01.000000Z"), Instant.parse("2038-01-19T03:14:07.499999Z"))));


Why not 1970-01-01T00:00:00.000000Z?

https://en.wikipedia.org/wiki/Unix_time

Unix time is currently defined as the number of seconds which have passed since 00:00:00 UTC on Thursday, 1 January 1970

good catch, this should proably be 1970-01-01T00:00:00.000001Z - i.e. epoch + min delta

From the MySQL documentation:

The TIMESTAMP data type is used for values that contain both date and time parts. TIMESTAMP has a range of '1970-01-01 00:00:01' UTC to '2038-01-19 03:14:07' UTC.

mysql> insert into test_unsupported_timestamp values ('1970-01-01 00:00:00'); ERROR 1292 (22007): Incorrect datetime value: '1970-01-01 00:00:00' for column 'data' at row 1 mysql> insert into test_unsupported_timestamp values ('1970-01-01 00:00:01'); Query OK, 1 row affected (0.01 sec) mysql> select * from test_unsupported_timestamp; +---------------------+ | data | +---------------------+ | 1970-01-01 00:00:01 | +---------------------+

I'm not sure whether this is a problem or not, but I'll mention it 'just in case'.
Back a few years ago I was working with a JSON interface of an HTTP endpoint. I was surprised to see that I could convert a negative long passed in JSON into Javascript date to represent dates before epoch. We used that for birth dates just like described here: https://www.epochconverter.com/programming/mysql (paragraph 'Negative epochs').

Now, as the article reads, mysql has a problem representing timestamps before epoch. However, we have a 'converter' at hand that is translating trino datatype to particular database datatype. And that trino datatype just might support negative values. Currently the test throws an exception if the timestamp is out of range, for instance if it's before epoch. I'm just wondering whether that's the behaviour we want to have. Maybe we somehow want to emulate negatives (ie. before epoch) timestamps.

Treat it as a general comment that can be ignored.

No. That makes such "emulated" values only readable by Trino. It's a form of vendor lock in.

All connectors have tests (Test*TypeMapping) which ensure that values written by trino round trip when read directly from the db and the other way around as well.

@adamjshook please add a code comment about the min value to make it more obvious

dominikzalewski · 2023-08-24T11:46:22Z

plugin/trino-mysql/src/main/java/io/trino/plugin/mysql/TrinoToMySqlWriteChecker.java

+        @Override
+        public void validate(Instant value)
+        {
+            if (value.isBefore(range.getMin()) || value.isAfter(range.getMax())) {


In other places in the project (like io.trino.plugin.jdbc.JdbcMetadataConfig), I see we seem to be using a validation framework, for instance jakarta.validation.constraints.Min. Is there a reason (performance?) we cannot use it here as well?

performance, jakarta validation does too much work. It's more useful for user-facing validation than runtime checks.

dominikzalewski · 2023-08-24T11:59:06Z

plugin/trino-mysql/src/main/java/io/trino/plugin/mysql/TrinoToMySqlWriteChecker.java

+        }
+    }
+
+    private static class Range<T>


There is a range class in apache commons:

https://commons.apache.org/proper/commons-lang/apidocs/org/apache/commons/lang3/Range.html

That already has most of the logic introduced here. I see that we have commons in the master pom.xml:

<dependency> <groupId>org.apache.commons</groupId> <artifactId>commons-lang3</artifactId> <version>3.13.0</version> </dependency>

Is there a reason why we cannot use it in plugins?

dominikzalewski · 2023-08-24T12:08:49Z

plugin/trino-mysql/src/main/java/io/trino/plugin/mysql/TrinoToMySqlWriteChecker.java

+import static java.lang.String.format;
+import static java.util.Objects.requireNonNull;
+
+public class TrinoToMySqlWriteChecker<T>


I don't think I understand the complexity introduced with these abstractions. The only two production usages of the validate() method I can see is that we pass an Instance and the code checks under the covers if it's within epoch's min and max values. Do we really need to introduce private interface, private class, another private class for Range in order to do that?

It seems that the epoch min and epoch max are hardcoded anyway. Isn't it the case that the only thing we need is a

private static boolean validate(Instant i) { return i >= Instant.parse("1970-01-01T00:00:00.000000Z") && Instant.parse("2038-01-19T03:14:07.499999Z") }

(or something like above that compiles or throws an exception instead of returning a boolean) in MySqlClient class?

I would understand wrapping into classes, if the code was unit-tested. But it does not look like there's a unit test for this class?

good point, we only have the TIMESTAMP validator here. ClickHouse on the other hand (where this class originates from) has 6 validators, each of them having different "rules" depending on the ClickHouse version we are connecting to.

Agreed this could be simplified into a static validateMySqlTimestampValue function in MySqlClient.

dominikzalewski · 2023-08-24T12:42:18Z

plugin/trino-mysql/src/main/java/io/trino/plugin/mysql/MySqlClient.java

+            throw new TrinoException(NOT_SUPPORTED, "This connector does not support creating tables with column comment");
+        }
+
+        return "%s %s %s" .formatted(


the space after the string here is intentional?

No, for some reason my IntelliJ formatter put that there. I'll fix it.

dominikzalewski · 2023-08-24T12:45:13Z

plugin/trino-mysql/src/main/java/io/trino/plugin/mysql/MySqlClient.java

@@ -469,6 +501,13 @@ public Optional<ColumnMapping> toColumnMapping(ConnectorSession session, Connect
                return Optional.of(jsonColumnMapping());
            case "enum":
                return Optional.of(defaultVarcharColumnMapping(typeHandle.getRequiredColumnSize(), false));
+            case "datetime":
+                TimestampType timestampType = createTimestampType(getTimestampPrecision(typeHandle.getRequiredColumnSize()));


closing this logic into a separate private method that would be named like 'mysqlDateTimeToTrinoTimestamp' would make it more explicit for the reader that this is what's going on here.

dominikzalewski · 2023-08-24T12:46:50Z

plugin/trino-mysql/src/main/java/io/trino/plugin/mysql/MySqlClient.java

-                        timestampType,
-                        mySqlTimestampReadFunction(timestampType),
-                        timestampWriteFunction(timestampType)));
+                TimestampWithTimeZoneType trinoType = createTimestampWithTimeZoneType(getTimestampPrecision(typeHandle.getRequiredColumnSize()));


similarly here 'mysqlTimestampToTrinoTimestampWithTZ'

dominikzalewski · 2023-08-24T12:50:42Z

plugin/trino-mysql/src/main/java/io/trino/plugin/mysql/MySqlClient.java

+                    OffsetDateTime offsetDateTime = resultSet.getObject(columnIndex, OffsetDateTime.class);
+                    return LongTimestampWithTimeZone.fromEpochSecondsAndFraction(
+                            offsetDateTime.toEpochSecond(),
+                            (long) offsetDateTime.getNano() * PICOSECONDS_PER_NANOSECOND,


for super-easy reading maybe wrap this computation into a private method called 'nanos2pico'?

The read and write functions are modeled after a few existing functions. They could be put in one place and shared but that feels out of scope of this change and could be resolved as a separate issue.

dominikzalewski · 2023-08-24T12:54:04Z

plugin/trino-mysql/src/main/java/io/trino/plugin/mysql/MySqlClient.java

@@ -592,6 +636,50 @@ private static ColumnMapping mySqlCharColumnMapping(CharType charType, Optional<
        return ColumnMapping.sliceMapping(charType, charReadFunction(charType), charWriteFunction(), pushdownController);
    }

+    private static LongReadFunction shortTimestampWithTimeZoneReadFunction()


I'm not sure whether the 'short' and 'long' prefix in the following 4 methods is not misleading. The reason being that the returned type has sometimes 'Long' in the name, so I had to read it 4 times and browse deep into the code to understand why is that. Maybe there is a better wording like for instance 'simpleTimestamp'/'complexTimestamp' or 'regularTimestamp'/'extendedTimestamp'?

See #18470 (comment) -- similar answer for why they are named the way they are.

The reason is even more than that. The trino classes which implement the type are called ShortTimestampType and LongTimestampType and that's where the name comes from.

dominikzalewski · 2023-08-24T12:57:58Z

plugin/trino-mysql/src/main/java/io/trino/plugin/mysql/MySqlClient.java

+        return ObjectWriteFunction.of(
+                LongTimestampWithTimeZone.class,
+                (statement, index, value) -> {
+                    long epochSeconds = floorDiv(value.getEpochMillis(), MILLISECONDS_PER_SECOND);


I would wrap it into a private method called 'epochMillis2Seconds'
and the below into 'timestamp2nanosDelta'.

Then I'd inline it:

Instant instantValue = Instant.ofEpochSecond(epochMillis2Seconds(value), timestamp2nanosDelta(value))

adamjshook · 2023-08-24T13:04:53Z

@adamjshook Does mysql.md (the docs) need to be updated as well to mention the new mapping?

Yes! I'll add that as a separate commit when making other updates to this PR and we'll need a note in the release notes as well regarding the change.

dominikzalewski · 2023-08-24T15:15:32Z

...trino-mysql/src/test/java/io/trino/plugin/mysql/TestMySqlTimeMappingsWithServerTimeZone.java

+    @Test(dataProvider = "sessionZonesDataProvider")
+    public void testDate(ZoneId sessionZone)
+    {
+        Session session = Session.builder(getSession())


this snippet is repeated in pratically every test. Do you think there's value in promoting it to an instance variable and initializing in a setup() method?

dominikzalewski · 2023-08-24T15:19:20Z

In the PR description it reads:

Note that MySQL DATETIME types are read as TIMESTAMP types and Trino TIMESTAMP types are inserted into MySQL as DATETIME types -- this behavior is unchanged.

Due to the size of the change (on test side), it's hard for me to verify whether this condition is met, or whether we need the backward compatibility test for that, etc. Having written that I just want to make sure that this has been thought through, even if my ask is 'doubled effort' since this has already been addressed.

adamjshook · 2023-08-24T15:33:35Z

I'm trying to dig in to why we need this checker in the first place. I find it a bit unsettling that you insert a java.sql.Timestamp initialized with UTC milliseconds and the MySQL server feels like it needs to convert it to the server time zone. Particularly when we explicitly set connectionTimeZone=UTC in the MySqlClientModule.

// Trino
TIMESTAMP '2019-03-18 10:01:17.987 UTC

// MySQL
mysql> SET time_zone = 'UTC';
mysql> select * from test_timestamp_with_time_zone613fa1muom;
+-------------------------+
| col_0                   |
+-------------------------+
| 2019-03-17 20:01:17.987 |
+-------------------------+

dominikzalewski · 2023-08-24T15:36:52Z

...trino-mysql/src/test/java/io/trino/plugin/mysql/TestMySqlTimeMappingsWithServerTimeZone.java

+
+        SqlDataTypeTest.create()
+                .addRoundTrip("date", "DATE '0001-01-01'", DATE, "DATE '0001-01-01'")
+                .addRoundTrip("date", "DATE '1582-10-04'", DATE, "DATE '1582-10-04'") // before julian->gregorian switch


maybe it would be beneficial to have methods named 'beforeJulianGregorianSwitch' in SqlDataTypeTest class. This was the code would document itself, would be more readable and maintainable (as comment would be unnecessary)

dominikzalewski · 2023-08-24T15:37:35Z

...trino-mysql/src/test/java/io/trino/plugin/mysql/TestMySqlTimeMappingsWithServerTimeZone.java

+    }
+
+    @Test(dataProvider = "sessionZonesDataProvider")
+    public void testTimeFromMySql(ZoneId sessionZone)


feels like this test should be named 'testPrecision'?

dominikzalewski · 2023-08-24T15:43:22Z

...trino-mysql/src/test/java/io/trino/plugin/mysql/TestMySqlTimeMappingsWithServerTimeZone.java

+
+        SqlDataTypeTest.create()
+                // default precision in MySQL is 0
+                .addRoundTrip("TIME", "TIME '00:00:00'", createTimeType(0), "TIME '00:00:00'")


These literals seem to be repeated throughout the test. Do you think there's value extracting them to a constant and reusing?

adamjshook · 2023-09-02T17:23:55Z

From here solution 2b seems to be the one that is working best. Setting the below properties results in the value stored in MySQL to accurately reflect the value inserted via Trino. I've removed code that validates the timestamptz values since MySQL is returning an expected error now.

connectionProperties.setProperty("connectionTimeZone", "LOCAL");
connectionProperties.setProperty("forceConnectionTimeZoneToSession", "true");

// Trino
TIMESTAMP '2019-03-18 10:01:17.987 UTC

mysql> set time_zone = utc;
mysql> select * from test_timestamp_with_time_zone;
+-------------------------+
| col_0                   |
+-------------------------+
| 2019-03-18 10:01:17.987 |
+-------------------------+

One thing to note is the TestMySqlTimeMappingsWithServerTimeZone#testTimestampFromMySql test case where MySQL TIMESTAMP values are inserted with the MySQL JDBC driver. The time zones of these are set to the server time zone, Pacific/Apia, and then read back in UTC as expected. I think this behavior is fine and expected, but I wanted to call it out.

hashhar

@ebyhr / @findepi / @anusudarsan can one of you also please take a look?

hashhar · 2023-09-08T08:20:58Z

plugin/trino-mysql/src/main/java/io/trino/plugin/mysql/MySqlClient.java

@@ -208,6 +225,10 @@ public class MySqlClient
    // MySQL driver returns width of time types instead of precision, same as the above timestamp type.
    private static final int ZERO_PRECISION_TIME_COLUMN_SIZE = 8;

+    // MySQL TIMESTAMP has a range of '1970-01-01 00:00:01' UTC to '2038-01-19 03:14:07' UTC
+    private static final Instant MYSQL_TIMESTAMP_MIN_VALUE = Instant.parse("1970-01-01T00:00:01.000000Z");
+    private static final Instant MYSQL_TIMESTAMP_MAX_VALUE = Instant.parse("2038-01-19T03:14:07.499999Z");


that .49999 is confusing me, can you explain me why? It's > 03:14:07 UTC.

The following paragraph from here (emphasis mine) refers to the fractional seconds. Kind of misleading as the previous paragraph says the maximum is '2038-01-19 03:14:07' UTC

A DATETIME or TIMESTAMP value can include a trailing fractional seconds part in up to microseconds (6 digits) precision. In particular, any fractional part in a value inserted into a DATETIME or TIMESTAMP column is stored rather than discarded. With the fractional part included, the format for these values is 'YYYY-MM-DD hh:mm:ss[.fraction]', the range for DATETIME values is '1000-01-01 00:00:00.000000' to '9999-12-31 23:59:59.499999', and the range for TIMESTAMP values is '1970-01-01 00:00:01.000000' to '2038-01-19 03:14:07.499999'. The fractional part should always be separated from the rest of the time by a decimal point; no other fractional seconds delimiter is recognized.

hashhar · 2023-09-08T08:23:45Z

plugin/trino-mysql/src/main/java/io/trino/plugin/mysql/MySqlClientModule.java

-        // TODO (https://github.com/trinodb/trino/issues/15668) rethink how timestamps are mapped. Also, probably worth adding tests
-        //  with MySQL server with a non-UTC system zone.
-        connectionProperties.setProperty("connectionTimeZone", "UTC");
+        // Force the connection time zone of that to the server to preserve time zones


sounds opposite of what MySQL docs say:

according to docs: connectionTimeZone = LOCAL means that JDBC driver uses the JVM zone as the session zone.
forceConnectionTimeZoneToSession = true means that the server side connection zone is changed to match local JVM zone.

Now in addition to this I started thinking about whether forceConnectionTimeZoneToSession = true can lead to failure if the JVM zone doesn't exist on the server. This probably explains the existing comment about without relying on server time zone (which may be configured to be totally unusable)..

see #15670 as an example of an issue I'm thinking about but in reverse direction.

Yes, you're right. Comment is now backwards; I didn't update it from a previous attempt to get this to work. I'll update it to reflect what is in the documentation.

hashhar

Thanks.

We can address any post-merge comments from @anusudarsan/@findepi if they take a look.

cla-bot bot added the cla-signed label Jul 31, 2023

adamjshook requested a review from hashhar July 31, 2023 22:45

adamjshook force-pushed the adamjshook/mysql-timestamps branch from 411146f to 062a6ed Compare August 1, 2023 18:15

hashhar reviewed Aug 2, 2023

View reviewed changes

adamjshook force-pushed the adamjshook/mysql-timestamps branch from 062a6ed to 90fe963 Compare August 3, 2023 13:57

adamjshook force-pushed the adamjshook/mysql-timestamps branch from 90fe963 to 1996d18 Compare August 23, 2023 18:15

hashhar approved these changes Aug 24, 2023

View reviewed changes

hashhar requested review from wendigo and ebyhr August 24, 2023 04:53

dominikzalewski reviewed Aug 24, 2023

View reviewed changes

hashhar reviewed Aug 24, 2023

View reviewed changes

dominikzalewski reviewed Aug 24, 2023

View reviewed changes

ebyhr removed their request for review August 30, 2023 02:10

adamjshook force-pushed the adamjshook/mysql-timestamps branch from 1996d18 to 8cd1e31 Compare September 2, 2023 17:43

adamjshook requested a review from hashhar September 2, 2023 17:43

github-actions bot added the docs label Sep 2, 2023

hashhar reviewed Sep 8, 2023

View reviewed changes

hashhar requested review from findepi, ebyhr and anusudarsan September 8, 2023 08:34

adamjshook added 2 commits September 12, 2023 13:31

Map MySQL TIMESTAMP types to TIMESTAMP WITH TIME ZONE

b43ac73

Update Trino and MySQL type mappings in connector docs

41cef73

adamjshook force-pushed the adamjshook/mysql-timestamps branch from 8cd1e31 to 41cef73 Compare September 12, 2023 13:32

hashhar approved these changes Sep 13, 2023

View reviewed changes

hashhar merged commit fd5b247 into trinodb:master Sep 13, 2023

github-actions bot added this to the 427 milestone Sep 13, 2023

colebow mentioned this pull request Sep 13, 2023

Add Trino 427 release notes #19023

Merged

adamjshook deleted the adamjshook/mysql-timestamps branch October 2, 2023 13:15

adamjshook mentioned this pull request Oct 2, 2023

Add docs for MySQL timestamptz handling #19231

Merged

		// after epoch (MySQL's timestamp type doesn't support values <= epoch)
		.addRoundTrip("timestamp(3) WITH TIME ZONE", "TIMESTAMP '2019-03-18 10:01:17.987 %s'".formatted(sessionZone), createTimestampWithTimeZoneType(3), "TIMESTAMP '2019-03-18 10:01:17.987 UTC'")

Map MySQL TIMESTAMP types to TIMESTAMP WITH TIME ZONE #18470

Map MySQL TIMESTAMP types to TIMESTAMP WITH TIME ZONE #18470

Conversation

adamjshook commented Jul 31, 2023 • edited by hashhar Loading

Description

Additional context and related issues

Release notes

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

adamjshook commented Aug 3, 2023

adamjshook commented Aug 23, 2023

hashhar commented Aug 23, 2023

hashhar commented Aug 24, 2023

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

hashhar left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

adamjshook Aug 24, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

adamjshook commented Aug 24, 2023

Choose a reason for hiding this comment

dominikzalewski commented Aug 24, 2023

adamjshook commented Aug 24, 2023

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

adamjshook commented Sep 2, 2023

hashhar left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

hashhar Sep 8, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

hashhar left a comment

Choose a reason for hiding this comment

adamjshook commented Jul 31, 2023 •

edited by hashhar

Loading

adamjshook Aug 24, 2023 •

edited

Loading

hashhar Sep 8, 2023 •

edited

Loading