[GROOVY-10909] bytecode for in, !in, !() and ?:

• Type: Improvement
• Status: Closed
• Priority: Minor
• Resolution: Incomplete
• Fix Version/s: 5.0.0-alpha-1
• Component/s: bytecode, Compiler, Static compilation
• Labels: bytecode
• Assignee: Eric Milles
• Reporter: Eric Milles
• Created: 20/Jan/23 14:52
• Updated: 01/Feb/23 23:31
• Resolved: 01/Feb/23 23:30

Description

Consider the following:

List getList() { ['xx'] }
@groovy.transform.CompileStatic
void test(object) {
  def others = [], strings = []
  if (object in list) { // 4 of 8 branches missed
    strings << object
  }
  if (!(object in list)) { // 4 of 10 branched missed
    others << object
  }
  println others
  println strings
}
test(null)
test('xx')
test([''])

When unit testing, I noticed 4 of 10 branches missed for "if (!(item in list)) ...". When compiled and executed dynamically, it shows 2 of 2 branches covered (safe call to "list.isCase(item)").

Static compilation transforms "item in list" into "list == null ? item == null : list.isCase(item)". This is certainly where all the branches are coming from.

Issue Links

• is related to: GROOVY-7473 inefficient code generation
• is related to: GROOVY-10383 SC: !in compiled to ScriptBytecodeAdapter#isNotCase
• relates to: GROOVY-10915 SC: class that provides isCase but not isNotCase