Sum of empty t-digest should be null #138378
Merged
+20
−10
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
not-napoleon
added
>non-issue
test-release
Collaborator
|
Pinging @elastic/es-storage-engine (Team:StorageEngine) |
42 tasks
kkrik-es
approved these changes
Nov 21, 2025
Member
Author
|
I'm removing the release test tag as this does not appear to introduce any new release test failures, and there are several unrelated issues blocking merging. |
szybia
added a commit
to szybia/elasticsearch
that referenced
this pull request
Nov 24, 2025
…-json * upstream/main: (247 commits) Mute org.elasticsearch.xpack.inference.integration.SemanticTextIndexOptionsIT testValidateIndexOptionsWithBasicLicense elastic#138513 Mute org.elasticsearch.xpack.esql.heap_attack.HeapAttackLookupJoinIT testLookupExplosionBigString elastic#138510 This shouldn't be zero (elastic#138501) sum of empty histogram is now null (elastic#138378) Test ES|QL bfloat16 support (elastic#138499) Fix exception handling in S3 `compareAndExchangeRegister` (elastic#138488) Mute org.elasticsearch.xpack.exponentialhistogram.ExponentialHistogramFieldMapperTests testFormattedDocValues elastic#138504 Mute org.elasticsearch.ingest.geoip.IngestGeoIpClientYamlTestSuiteIT test {yaml=ingest_geoip/60_ip_location_databases/Test adding, getting, and removing ip location databases} elastic#138502 ESQL: Refactor HeapAttackIT (elastic#138432) [Inference API] Add ElasticInferenceServiceDenseTextEmbeddingsServiceSettings to InferenceNamedWriteablesProvider (elastic#138484) Store split indices (elastic#138396) ES|QL Update CHUNK to support chunking_settings as optional argument (elastic#138123) Extract common blob-update logic in `S3HttpHandler` (elastic#138490) Cleanup esql request building api (elastic#138398) Round sum and avg in exponential_histogram CSV tests (elastic#138472) ESQL: load exponential_histogram total count as double instead of long (elastic#138417) [SIMD] Use fixed width native types for better Java interoperability (elastic#138429) Do not use Min or Max as Top's surrogate when there is an outputField (elastic#138380) ES|QL: Fix generative tests (elastic#138478) Mute org.elasticsearch.xpack.inference.integration.AuthorizationTaskExecutorIT testCreatesEisChatCompletion_DoesNotRemoveEndpointWhenNoLongerAuthorized elastic#138480 ...
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Part of #137988
Implement #138349 for the t-digest field type.
This changes the behavior of the sum sub-field of t-digest when the digest is empty. Prior to this PR we treated it as 0, and this changes it to be null. This avoids a division by zero error in ESQL when trying to calculate the average of an empty histogram. We are adopting the same behavior for the exponential histogram field (see PR linked above), and this is important to keep the semantics of the two fields as close as possible.
I've put the release tests tag on this to confirm that it doesn't break anything, but the only impacted tests should already be skipped in BWC so I don't expect to need to change the skip logic.