Nhat Nguyen

> I'm not sure about this -- even if all indices use ECS I think there are often differences (some index has one field, another is missing it, etc.)? I'm...

@jtibshirani I've pushed a general optimization for the merging process. The new optimization reduces the response time significantly (9-10 times) for field-caps requests targeting single or multiple index mappings. Can...

@jtibshirani @javanna I've revised this PR. Can you please take another look? > This would also avoid the need to rely on map object equality as we do in the...

> Is the comparison using getIndexMappingHash much slower than the object equality comparison? I agree with @jtibshirani, this feels a bit fragile as it relies on the deserialization code elsewhere...

@jtibshirani @romseygeek @Leaf-Lin thank you for reviews.

Before this change, I took a different approach, creating a [MetricsAggregationOperator](https://gist.github.com/dnhatn/b48d36cc9d3e93edc6f325e635e01333) that extends the HashAggregationOperator and overrides the output of the final mode. However, I believe this current approach is...

test this please

@kkrik-es @martijnvg @nik9000 Thank you for your review and feedback. I am going to merge this PR as is. I think we will need some adjustments when integrating with the...

Thanks @davemoore- for reporting this. @jtibshirani your explanation is correct. This test creates a single segment (the dataset is quite small), and TierMergePolicy thinks that segment [is merged already](https://github.com/apache/lucene-solr/blob/425f207f4098e4fcc1307729b1f2e17cfef3929f/lucene/core/src/java/org/apache/lucene/index/TieredMergePolicy.java#L701). Hence,...

Thanks Alex!