zentol commented on a change in pull request #12682:
URL: https://github.com/apache/flink/pull/12682#discussion_r441451360
##########
File path:
flink-connectors/flink-sql-connector-hive-1.2.2/src/main/resources/META-INF/NOTICE
##########
@@ -8,6 +8,41 @@ This project bundles the following dependencies under the
Apache Software Licens
- org.apache.hive:hive-exec:1.2.2
- org.apache.hive:hive-metastore:1.2.2
+- org.apache.hive:hive-common:1.2.2
+- org.apache.hive:hive-serde:1.2.2
+- org.apache.hive.shims:hive-shims-0.20S:1.2.2
+- org.apache.hive.shims:hive-shims-0.23:1.2.2
+- org.apache.hive.shims:hive-shims-common:1.2.2
+- org.apache.hive:spark-client:1.2.2
+- com.twitter:parquet-hadoop-bundle:1.6.0
+- org.apache.thrift:libthrift:0.9.2
- org.apache.thrift:libfb303:0.9.2
- org.apache.orc:orc-core:1.4.3
- io.airlift:aircompressor:0.8
+- commons-lang:commons-lang:2.6
+- org.apache.commons:commons-lang3:3.1
+- org.apache.avro:avro:1.7.5
+- org.apache.avro:avro-mapred:1.7.5
+- com.googlecode.javaewah:JavaEWAH:0.3.2
+- org.iq80.snappy:snappy:0.2
+- org.codehaus.jackson:jackson-core-asl:1.9.2
+- org.codehaus.jackson:jackson-mapper-asl:1.9.2
+- com.google.guava:guava:14.0.1
+- net.sf.opencsv:opencsv:2.3
+- joda-time:joda-time:2.5
+- org.objenesis:objenesis:1.2
+
+This project bundles the following dependencies under the BSD license.
+See bundled license files for details.
+
+- com.esotericsoftware.kryo:kryo:2.22
+- org.jodd:jodd-core:3.5.2
+- javolution:javolution:5.5.1
+- com.google.protobuf:protobuf-java:2.5.0
+- com.esotericsoftware.minlog:minlog:1.2
+- com.esotericsoftware.reflectasm:reflectasm:1.07
+
+This project bundles the following dependencies under the JSON license.
Review comment:
>I believe we can keep Hive 1.x support in the source code, with the bad
JSON dependency, if we add a warning to the documentation.
We did not only did not provide a binary license, but also had an opt-in
profile for including the module.
Technically this is also the case right now (since we have a dedicated
profile for setting the version to hive 1.1.x), but IMO it's not distinct
enough from the other profiles. It very much looks like "just another" a
version switch.
With that in mind, we'd either need to splice out hive 1.1 from the existing
hive-connector (not feasible for practical reason I suppose), have an opt-in
profile for hive as a whole (likely not feasible since the entire table
ecosystem is interlinked like a web), or make the hive 1.1 profile especially
explicit (in the documentation and source).
In any case however nothing we release to maven central or bundle in the
distribution must have a reference to hive 1.1.
This implies that users would have to recompile Flink against hive 1.1 to
use it. Given the reported frictions with Flink users getting up and running in
the Table/SQL APIs in regards to connectors, which is especially worse with
Hive, this to me implies that it would be basically unusable. IIRC, the
_current state_ was already described as such.
Hence why I referred to this as not being _feasible_. _Possible_, yes, but I
daresay that it's not worth the effort.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
