FrankChen021 opened a new pull request, #19688:
URL: https://github.com/apache/druid/pull/19688

   ### Description
   
   Upgrade RoaringBitmap from 1.6.13 to 1.6.14.
   
   RoaringBitmap 1.6.13 invalidates its `Roaring64NavigableMap` 
cumulative-cardinality cache using signed `Math.min` even when high keys are 
ordered as unsigned values. Merging a bitmap containing a negative `long` value 
with one containing non-negative values can therefore leave the cache in an 
inconsistent state and fail while reading cardinality.
   
   RoaringBitmap fixed this in [issue 
#828](https://github.com/RoaringBitmap/RoaringBitmap/issues/828) and [PR 
#829](https://github.com/RoaringBitmap/RoaringBitmap/pull/829), released in 
1.6.14. This PR also adds a regression test through Druid's 
`Bitmap64ExactCountMergeAggregator` path.
   
   Before the dependency upgrade, the regression test fails with:
   
   ```
   java.lang.AssertionError: 1 is bigger than 0
     at org.roaringbitmap.longlong.Roaring64NavigableMap.ensureOne(...)
     at org.roaringbitmap.longlong.Roaring64NavigableMap.getLongCardinality(...)
   ```
   
   With 1.6.14, the regression and all 78 `Bitmap64ExactCount*Test` tests pass.
   
   #### Release note
   
   Upgrades RoaringBitmap to 1.6.14 to prevent bitmap64 exact-count queries 
from failing when merged bitmaps span negative and non-negative `long` values.
   
   <hr>
   
   ##### Key changed/added classes in this PR
    * `Bitmap64ExactCountMergeAggregatorTest`
   
   <hr>
   
   This PR has:
   
   - [x] been self-reviewed.
   - [x] a release note entry in the PR description.
   - [x] added or updated version, license, or notice information in 
`licenses.yaml`.
   - [x] added unit tests to cover the affected merge path.
   


-- 
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]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to