Repository navigation
Build a typed temporal aggregate without a Spark Column - #158
Merged
estebanzimanyi merged 1 commit intoOct 5, 2026
Merged
Conversation
The builder that the spark-sql engine emits for a temporal aggregate of the typed surface made its expression through the public Column API: it applied functions.udaf to new Column(argument) and read the expression back with Column.expr(). Spark 4 removes both, the Column of Spark 4 carrying a ColumnNode rather than an Expression, so the generated TemporalAggregates does not compile against Spark 4. The builder now makes the Catalyst aggregate itself, as the scalar builder of MeosSqlRuntime makes a ScalaUDF: a ScalaAggregator over the argument expressions, the Typed aggregator and the expression encoders of its input and its buffer, turned into an aggregate expression. The constructor and encoderFor have the same signatures in Spark 3.5 and Spark 4, so the generated class compiles against both, and MobilitySpark moves to Spark 4 without a window in which its build and this generator disagree. Witness: with the generator of main, MobilitySpark with Spark 4.0.1 fails to compile TemporalAggregates, "no suitable constructor found for Column(Expression)". Measured: MobilitySpark main regenerated with this generator against MobilityDB 7128c9344a passes its suite, 34 tests and no build warning, with Spark 3.5.1 and with Spark 4.0.1, the temporal aggregates of GeneratedSqlSurfaceTest among them. The codegen tests pass, 106 tests.
estebanzimanyi
added a commit
to MobilityDB/MobilitySpark
that referenced
this pull request
Oct 5, 2026
The pom moves spark-core and spark-sql from 3.5.1 to 4.0.1, and the job of the workflow names Spark 4.0. Spark 4.0 is the release the book's chapters on Spark run with Apache Sedona 1.8 and Delta Lake 4.0, whose environment rebuilt MobilitySpark against Spark 4.0 by editing this pom. The generated surface compiles against Spark 4 since MobilityDB/JMEOS#158, which builds a typed temporal aggregate without a Spark Column: before it, the generator applied functions.udaf to new Column(Expression), which Spark 4 removes, and TemporalAggregates did not compile. Measured: tools/refresh-from-master.sh from scratch, against MobilityDB 270106028b and JMEOS main 53ca034, passes the suite, 34 tests, with no build warning.
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
The builder that the spark-sql engine emits for a temporal aggregate of the
typed surface made its expression through the public Column API: it applied
functions.udaf to new Column(argument) and read the expression back with
Column.expr(). Spark 4 removes both, the Column of Spark 4 carrying a
ColumnNode rather than an Expression, so the generated TemporalAggregates
does not compile against Spark 4.
The builder now makes the Catalyst aggregate itself, as the scalar builder of
MeosSqlRuntime makes a ScalaUDF: a ScalaAggregator over the argument
expressions, the Typed aggregator and the expression encoders of its input
and its buffer, turned into an aggregate expression. The constructor and
encoderFor have the same signatures in Spark 3.5 and Spark 4, so the
generated class compiles against both, and MobilitySpark moves to Spark 4
without a window in which its build and this generator disagree.
Witness: with the generator of main, MobilitySpark with Spark 4.0.1 fails to
compile TemporalAggregates, "no suitable constructor found for
Column(Expression)".
Measured: MobilitySpark main regenerated with this generator against
MobilityDB 7128c9344a passes its suite, 34 tests and no build warning, with
Spark 3.5.1 and with Spark 4.0.1, the temporal aggregates of
GeneratedSqlSurfaceTest among them. The codegen tests pass, 106 tests.