Skip to content

Conversation

iverase
Copy link
Contributor

@iverase iverase commented Mar 30, 2021

For now this is just a copy of Lucene86PointsFormat. The existing Lucene86PointsFormat is moved to backwards-codecs.

This is actually the last codec that needs to be moved to Lucene90 :)

@iverase iverase requested a review from jtibshirani March 30, 2021 08:35
Copy link
Member

@jtibshirani jtibshirani left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

As you mentioned, this is the only codec we're moving that still shares a chunk of logic with older codecs (through BKDWriter and BKDReader). A couple questions:

  • How does this affect your planned work in LUCENE-9047, which was a big motivation for creating these new codecs? Will you just add another 'internal version' for the endianness change?
  • Confirming my understanding, we plan to follow up with something like LUCENE-9820 to avoid having to share so many classes with older codecs in the future?

super.testMergeStability();
}

public void testEstimatePointCount() throws IOException {
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is it possible to move these methods to BasePointsFormatTestCase? It'd be nice to cut down on duplicated test cases.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I moved some test but the ones that relay on knowing the maxPointsInLeafNode cannot be moved as the information is not expose through the PointValues API.

writeState.segmentInfo.name,
writeState.segmentSuffix,
Lucene86PointsFormat.DATA_EXTENSION);
Lucene86RWPointsFormat.DATA_EXTENSION);
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Small comment, seems clearer to reference Lucene86PointsFormat here.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

good catch

@iverase
Copy link
Contributor Author

iverase commented Mar 31, 2021

How does this affect your planned work in LUCENE-9047, which was a big motivation for creating these new codecs? Will you just add another 'internal version' for the endianness change?

I don't think we will need to add a new internal version. I think for backwards codecs we can wrap the IndexInputs so it would be transparent for the BKD reader.

Confirming my understanding, we plan to follow up with something like LUCENE-9820 to avoid having to share so many classes with older codecs in the future?

I hope so :). What I am trying to avoid is to have to rewrite many BKD readers if we decide to change the Points API (for example LUCENE-9619). So separating reading a specific Index format from the API should help.

@iverase iverase merged commit 8c9b954 into apache:main Apr 1, 2021
@iverase iverase deleted the Lucene86PointsFormat branch April 1, 2021 05:04
gsmiller pushed a commit to gsmiller/lucene that referenced this pull request Apr 1, 2021
janhoy pushed a commit to cominvent/lucene that referenced this pull request May 12, 2021
* Solr's GlobalTracer is gone. Use OpenTracing's GlobalTracer, or even better: getters on CoreContainer, SolrQueryRequest, etc.
* No more "samplingThreshold" cluster property.  Instead, configure sampling specific to the plugin.
* Ensure the active span is visible from all threads by using ExecutorUtil
* JaegerTracerConfigurator is now configured via sys props / env vars
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants