uschindler commented on a change in pull request #2176:
URL: https://github.com/apache/lucene-solr/pull/2176#discussion_r551208675
##########
File path: gradle/defaults-java.gradle
##########
@@ -60,3 +60,14 @@ allprojects {
}
}
}
+
+configure(project(":lucene:core")) {
Review comment:
To come back to the MR-JAR: Yes we may need it, but it can be solved
without, too.
The IndexInputs are completely separate (so they can life side-by side). The
changes to MMapDircetory are mainly a change from int to long for the chunk
size (this is why I also not yet removed the methods to query unmapping status,
thy would always return true for JDK>xxxx.
When we would have this productive, MMapDirectory would just return one of
the IndexInputs that fits your JDK. We may only need to move the code that maps
a file to ByteBuffer or MemorySegment to a separate class or maybe the
ByteBuffer-/MemorySegmentIndexInput class. Then we have 2 classes staying next
to each other, one for JDK 11 and one for JDK 16/17/18+, MMapDirectory just
chooses the correct one. Then we won't need a MR JAR, only some tricks to
compile a few classes with later `--release`.
With compilation we can use `--release` and of course a newer Gradle
version. Like Elasticsearch, people who want to compile Lucene/Solr would need
JDK 17, but the code is working for JDK 11+. Elasticsearch does exactly the
same: It is tested and compiled for Java 8, but to build it you need JDK 15.
----------------------------------------------------------------
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
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
