Repository navigation
UTF-16 scan test leaves its fixture open on Windows #8795
Description
What happened?
The encoding regression calls exhausted next() without observing hasNext == false, which closes the lazy reader. Windows then rejects deleting the still-open temporary file. Correct decoding should not fail during cleanup.
How to reproduce?
On Windows with JDK 17, run:
sbt "WorkflowOperator/testOnly org.apache.texera.amber.operator.source.scan.file.FileScanSourceOpDescSpec"
The unchanged suite records 13 passes and one cleanup failure. Related to the encoding test introduced in #8605.
Version/Branch
1.4.0-incubating-SNAPSHOT (main)
Activity