voonhous commented on code in PR #18837:
URL: https://github.com/apache/hudi/pull/18837#discussion_r3466870756
##########
pom.xml:
##########
@@ -130,6 +129,12 @@
<hive.avro.version>1.11.4</hive.avro.version>
<presto.version>0.273</presto.version>
<trino.version>390</trino.version>
+ <!-- Trino SPI version that hudi-trino-plugin main code compiles against.
-->
+ <trino.connector.version>481</trino.connector.version>
+ <!-- Trino version used by the hudi-trino-tests profile for *-tests.jar
classifier deps.
+ Trino does not publish trino-spi / trino-filesystem / trino-hive
test-jars; the
+ profile is opt-in and assumes a locally-built Trino snapshot in
~/.m2. -->
+ <trino.connector.test.version>481</trino.connector.test.version>
Review Comment:
`trino.version` (390) is still used, just for a different thing -- it's the
`io.trino:trino-jdbc` client version for `hudi-integ-test`'s `TrinoQueryNode`
(and it's shaded into `hudi-integ-test-bundle`), not the connector. The
connector can't reuse it: it targets Trino SPI 481 (JDK 25), so 390 is far too
old to compile against, which is why it carries its own
`trino.connector.version` / `trino.connector.test.version`. Removing it
outright would break the integ-test build, so I've left it as-is in this PR.
It is a real disparity though -- the integ-test query nodes are pinned to
stale JDBC clients (Trino `390` via `TrinoQueryNode`, Presto `0.273` via
`PrestoQueryNode`) that lag well behind the connector. I've filed #19059 to
bump both query nodes up to current versions; once `TrinoQueryNode` moves to
481 we can collapse `trino.version` into the connector property and drop the
split entirely.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
