Project membership question.

[image: Screenshot From 2026-08-13 17-26-00.png]
Hey all, you probably know me as "Edward Im going to die on this hill
Capriolo". I usually get myself knee deep in making something work that is
no longer the hip-in-thing. So today's adventure. I have a feature for the
Cassandra-spark connector it only supports APPEND mode and my agent
confidently thinks we can make UPDATE mode work..

https://github.com/apache/cassandra-spark-connector

This is the canonical repo "it seems?" but no updates for a year since it
was dropped over there.

There does exist a "cassandra analytics" project
https://issues.apache.org/jira/projects/CASSANALYTICS?selectedItem=com.atlassian.jira.jira-projects-plugin:components-page

However my agent (possibly wrong is telling me that project doesn't govern
the code that I would need to change in the connector. See below:

--------------------------------------------------------------------



You’re right to call that out. apache/cassandra-analytics is the
active Apache Cassandra analytics repo.

PR apache/cassandra-analytics#231 confirms it:

Opened Aug 10, 2026.
Tied to CASSANALYTICS-190.
Against apache/cassandra-analytics:trunk.
Recent active branch / CI.
Adds version/distribution override seams around Cassandra internals.
But for our exact problem:

apache/cassandra-analytics is the active future-looking repo.
It appears focused on bulk analytics, SSTable/Sidecar reads/writes,
CDC, bridge/version compatibility.
Its current Spark SQL sink path is not the same as the old row-level
Spark Cassandra connector.
I did not see StreamingWrite, WriteBuilder, OutputMode, or
SupportsStreamingUpdateAsAppend in its Java/Scala source.
So it does not currently look like a drop-in replacement for our
low-volume microbatch upsert sink.
So the corrected answer is:

Canonical active Cassandra analytics
repo:https://github.com/apache/cassandra-analytics

Canonical old row-level Spark Cassandra connector
repo:https://github.com/apache/cassandra-spark-connector

For the “update every micro-batch into Cassandra” work, the old
connector is still the shortest path because the change is one Spark
V2 marker trait on the existing direct row writer.

For a broader “this is where Cassandra Spark work is alive now”
effort, cassandra-analytics is the repo to watch or target.

ok search this repo for the update mode
I’ll search apache/cassandra-analytics specifically for Spark
Structured Streaming output/update-mode support and related Jira/PR
history.

No direct code/doc hits for Spark streaming update mode in the repo.
I’ll broaden to Spark DataSource sink/write classes to see what write
APIs this repo actually exposes.

I searched apache/cassandra-analytics for update-mode terms.

No Hits No matches for:

OutputMode
SupportsStreamingUpdateAsAppend
StreamingWrite
WriteBuilder
buildForStreaming
STREAMING_WRITE
writeStream
outputMode
Update mode
What The Repo Has It has a V1 batch bulk sink:

cassandra-analytics-core/src/main/java/org/apache/cassandra/spark/sparksql/CassandraDataSink.java

It implements:

DataSourceRegister, CreatableRelationProvider
It only supports:

SaveMode.Append
and throws for overwrite/other modes.

Short name:

cassandraBulkWrite
The tests all use batch DataFrameWriter.save() with .mode("append").

Conclusion cassandra-analytics currently has no Spark Structured
Streaming update-mode support. Its writer path is
bulk/SSTable/Sidecar-oriented, not the direct low-volume row upsert
streaming sink we need.

For our micro-batch update use case, the old
apache/cassandra-spark-connector still has the relevant Spark V2
StreamingWrite implementation.

--------------------------------------------------

Now, I can be wrong as I often am. But I am going to suggest that the
canonical source isn't getting updated we should re-incubate it. I would be
happy to lead the effort and clean this thing up as I also have USEFUL
features I wish to add.

Thanks,
Edward