feat: implement cast from whole numbers to binary format and bool to decimal

[coderfender]

Which issue does this PR close?

1. Add int (byte/short/int/long) → binary cast support for LEGACY mode only (per Spark)
2. Add boolean → decimal cast support
3. Update cast tests to use DSL instead of SQL for regular casts (except try mode since try_cast is only supported starting Spark 4x)
4. Skip try_cast tests for casts not supported by Spark SQL

Relavent spark code :
https://github.com/apache/spark/blob/f89fbd49c45045b44259341c6c73e30ee9c46dd0/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/Cast.scala#L682

(Note that only legacy mode routes the code here while we Try and ANSI are not supported)

Closes #.

Rationale for this change

1. Add new comet native cast functionality from int / long to binary
2. update test suite to run DSL code (since spark supports whole numbers -> binary in DSL ,mode only)
3. Add support from boolean to decimal (at various scale and precision levels)

What changes are included in this PR?

How are these changes tested?

1. CastTestSuite tests and additional tests to cast Boolean to various Decimal Types

[coderfender]

@parthchandra , @andygrove these are the changes to support int to binary cast and bool to decimal cast operation . I have also had to change the base test to make sure that we use DSL (since Int -> Binary casts are only supported in the DSL side of Spark and would raise an encoding exception if we use Spark SQL) . Also, Int -> Binary casts do not support Try or ANSI eval modes in Spark and I have followed the same convention here as well

[coderfender]

Fixed format issue with compatibility.md ( I initially didn't add it but turns out the builds are failing with git errors)

[codecov-commenter]

Codecov Report

❌ Patch coverage is 63.82979% with 17 lines in your changes missing coverage. Please review.
✅ Project coverage is 59.98%. Comparing base (f09f8af) to head (6419906).
⚠️ Report is 918 commits behind head on main.

Files with missing lines	Patch %	Lines
...scala/org/apache/comet/expressions/CometCast.scala	63.82%	0 Missing and 17 partials ⚠️

Additional details and impacted files

@@             Coverage Diff              @@
##               main    #3083      +/-   ##
============================================
+ Coverage     56.12%   59.98%   +3.85%     
- Complexity      976     1464     +488     
============================================
  Files           119      175      +56     
  Lines         11743    16171    +4428     
  Branches       2251     2685     +434     
============================================
+ Hits           6591     9700    +3109     
- Misses         4012     5114    +1102     
- Partials       1140     1357     +217     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[andygrove]

Thanks for adding int→binary and boolean→decimal cast support!

Issue: Dead code

The function canCastToBinary (lines 354-358 in CometCast.scala) is defined but never called anywhere. This appears to be unused code that should be removed.

private def canCastToBinary(fromType: DataType): SupportLevel = fromType match {
  case DataTypes.ByteType | DataTypes.ShortType | DataTypes.IntegerType | DataTypes.LongType =>
    Compatible()
  case _ => Unsupported(Some(s"Cast from BinaryType to $fromType is not supported"))
}

The actual logic for int→binary casts is correctly handled in the canCastFromByte, canCastFromShort, canCastFromInt, and canCastFromLong functions.

---

Otherwise, the implementation looks good:

• Correctly limits int→binary to LEGACY mode only (matching Spark)
• Boolean→Decimal implementation handles precision/scale correctly
• Tests use column data from parquet (not just literals)
• Compatibility docs updated
• ANSI and Try modes correctly disabled for int→binary tests

---

This review was generated with AI assistance.

[andygrove]

There is a test failure:

2026-01-14T23:17:19.7092172Z - cast StringType to DecimalType(2,2) *** FAILED *** (684 milliseconds)
2026-01-14T23:17:19.7092965Z   "[[CAST_INVALID_INPUT] The value '
2026-01-14T23:17:19.7100086Z 3' of the type "STRING" cannot be cast to "DECIMAL(2,2)" because it is malformed. Correct the value as per the syntax, or change its target type. Use `try_cast` to tolerate malformed input and return NULL instead. If necessary set "spark.sql.ansi.enabled" to "false" to bypass this error]." did not equal "[[NUMERIC_VALUE_OUT_OF_RANGE] 3 cannot be represented as Decimal(2, 2). If necessary set "spark.sql.ansi.enabled" to "false" to bypass this error, and return NULL instead]." (CometCastSuite.scala:1424)
2026-01-14T23:17:19.7102421Z   Analysis:
2026-01-14T23:17:19.7102696Z   "[[CAST_INVALID_INPUT] The value '
2026-01-14T23:17:19.7105340Z 3' of the type "STRING" cannot be cast to "DECIMAL(2,2)" because it is malformed. Correct the value as per the syntax, or change its target type. Use `try_cast` to tolerate malformed input and return NULL instead. If necessary set "spark.sql.ansi.enabled" to "false" to bypass this error]." -> "[[NUMERIC_VALUE_OUT_OF_RANGE] 3 cannot be represented as Decimal(2, 2). If necessary set "spark.sql.ansi.enabled" to "false" to bypass this error, and return NULL instead]."

[coderfender]

Thank you @andygrove ,I am working on the test failures (I believe there is a bug on the DSL side vs using Spark SQL)

[coderfender]

There is a mismatch in the error message thrown by spark vs comet for a cast test from StringType to DecimalType(2,2) . I will raise a new PR to fix that error message .

Steps to Reproduce :

  test("cast StringType to DecimalType(2,2) -test error message") {
    withSQLConf(CometConf.getExprAllowIncompatConfigKey(classOf[Cast]) -> "true") {
      println("testing with simple input")
      val values = Seq("       3").toDF("a")
      Seq(true, false).foreach(ansiEnabled =>
        castTest(values, DataTypes.createDecimalType(2, 2), testAnsi = ansiEnabled))
    }
  }

[coderfender]

cc : @andygrove , @parthchandra

[coderfender]

I also found that the tests are not running on truly deterministic inputs . For example this test is failing (if I disable / change inputs for a test before that) :

  test("cast StringType to DecimalType(38,10) high precision") {
    withSQLConf(CometConf.getExprAllowIncompatConfigKey(classOf[Cast]) -> "true") {
      // TODO fix for Spark 4.0.0
      assume(!isSpark40Plus)
      val values = gen.generateStrings(dataSize, numericPattern, 38).toDF("a")
      Seq(true, false).foreach(ansiEnabled =>
        castTest(values, DataTypes.createDecimalType(38, 10), testAnsi = ansiEnabled))
    }
  }

[0e31,0E-10]                                   [0e31,null]

[coderfender]

Related PR to fix test in decimal cast : #3248

[coderfender]

@andygrove , @parthchandra I rebased the feature with main and the CometCastSuite finishes successfully on my local machine. Please review and kickoff CI whenever you get a chance

[parthchandra]

lgtm. (minor comment)

[coderfender]

Thank you for the approval @parthchandra

[coderfender]

@andygrove , @parthchandra , could you please review or potentially merge the branch whenever you get a chance

[coderfender]

Fixed comment indentation issues

[coderfender]

Rebased with main and fixed merged conflicts

[andygrove]

Thanks @coderfender