[james-willis]

Did you read the Contributor Guide?

• Yes, I have read the Contributor Rules and Contributor Development Guide

Is this PR related to a ticket?

• Yes, and the PR name follows the format [GH-XXX] my subject. Closes Bump GraphFrames from 0.11.0 to 0.12.2 and use randomized contraction for DBSCAN #3298

What changes were proposed in this PR?

Bumps <graphframes.version> in the root pom.xml from 0.11.0 to 0.12.2.

Sedona's only GraphFrames usage is GraphFrame(...).connectedComponents in DBSCAN.scala. The one breaking change in the 0.12.x line is the low-level Pregel edge-attribute contract (edge columns must now be requested via requiredEdgeColumns), which Sedona does not use. Artifact coordinates are unchanged and 0.12.2 is published for all three combinations Sedona builds against: graphframes-spark3_2.12, graphframes-spark3_2.13, graphframes-spark4_2.13.

Scope reduced

This PR originally also switched DBSCAN's connected components to randomized_contraction, as proposed in #3298. That part has been dropped, because the algorithm cannot run on Sedona's DBSCAN at all.

RandomizedContraction maps results back with coalesce(col(COMPONENT), col(ID)), where COMPONENT is a Long and ID is the original vertex id. DBSCAN.dbscan always sets id to sha2(to_json(struct("*")), 256) — a hex String — even when the input already has an id column (which is renamed to __id). Spark coerces the two branches to a common type, casts the hash to BIGINT, and fails:

org.apache.spark.SparkNumberFormatException: [CAST_INVALID_INPUT] The value
'36e53dd410ea44fd0f989eac55c1d9f90118dbf637374aa4e8733be40f3128f5' of the type
"STRING" cannot be cast to "BIGINT" because it is malformed. SQLSTATE: 22018

Every DBSCAN case in GeoStatsSuite failed on Spark 4.0.2 with this. Upstream, randomized_contraction is only exercised by TestLDBCCases, whose graphs are explicitly LongType, so non-integral vertex IDs are untested there. Reported upstream; the algorithm switch can be revisited once that is fixed and released.

I have left #3298 open to track the switch. Happy to close it instead if you would rather track only the bump.

How was this patch tested?

Full CI matrix. No source changes — dependency version only.

Did this PR include necessary documentation updates?

• No, this PR does not affect any public API so no need to change the documentation.