icksa opened a new issue, #18258:
URL: https://github.com/apache/druid/issues/18258
### Affected Version
33.0.0
### Description
I've reproduced this using the 33.0.0 quickstart configuration with no
modifications. I'm getting the following exception when attempting to compute
the median of a Kll Doubles Sketch using query:
```
select APPROX_QUANTILE_DS("kll_sketch", 0.5, 256) from "kll_doubles"
```
```
Error: RUNTIME_FAILURE (OPERATOR)
class
org.apache.datasketches.kll.KllDirectDoublesSketch$KllDirectCompactDoublesSketch
cannot be cast to class org.apache.datasketches.quantiles.DoublesSketch
(org.apache.datasketches.kll.KllDirectDoublesSketch$KllDirectCompactDoublesSketch
and org.apache.datasketches.quantiles.DoublesSketch are in unnamed module of
loader 'app')
```
The type of the column in Druid is `COMPLEX<KllDoublesSketch>`
## Steps to reproduce:
I've created a simple base64 kll doubles sketch:
```
#!/usr/bin/env python3
from datasketches import kll_doubles_sketch
import base64
import json
sketch = kll_doubles_sketch(k=200)
for i in range(1000):
sketch.update(i)
sketch_bytes = sketch.serialize()
base64_encoded_sketch = base64.b64encode(sketch_bytes).decode('utf-8')
with open("kll_sketches.csv", "w") as f:
f.write("timestamp,kll_sketch_column\n")
f.write(f"2025-07-15T00:00:00Z,{base64_encoded_sketch}\n")
```
I ingested it into druid using SQL:
```
REPLACE INTO "kll_doubles" OVERWRITE ALL
WITH "ext" AS (
SELECT * FROM TABLE (
EXTERN(
'{"type":"local","files":["/tmp/kll_sketches.csv"]}',
'{"type":"csv","findColumnsFromHeader":true}'
)
) EXTEND (
"timestamp" VARCHAR,
"kll_sketch_column" VARCHAR
)
)
SELECT
TIME_PARSE(TRIM("timestamp")) as __time,
COMPLEX_DECODE_BASE64('KllDoublesSketch', "kll_sketch_column") as
kll_sketch
FROM "ext"
PARTITIONED BY ALL
```
Attempt to query:
```
select APPROX_QUANTILE_DS("kll_sketch", 0.5, 256) from "kll_doubles"
````
--
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]
