The latest mvn verify run for GawkCompatibilityIT shows a concentrated group of failures where Jawk still disagrees with gawk on array-versus-scalar behavior.
Observed impact:
- 18 failing Gawk compatibility cases in the latest run
- Representative cases: aadelete2, arrayparm, arryref3, arryref4, arryref5, aryprm1, aryprm2, aryprm3, aryprm4, aryprm5, aryprm6, aryprm7, aryscl, mdim6, prmarscl, scalar, sclforin, sclifin
- Current failure modes include diagnostics such as:
- cannot assign anything to an unindexed associative array
- Cannot evaluate an unindexed array
- Cannot use IDAst ... as a scalar
This is now more visible because arrays-of-arrays support landed in #438, but several gawk behaviors around passing, referencing, deleting, iterating, and coercing subarrays still do not line up.
Acceptance criteria:
- the representative cases above pass in GawkCompatibilityIT
- array parameters and subarray references follow gawk's scalar-versus-array rules
- using an array in scalar context, or a scalar as an array, produces gawk-compatible behavior and diagnostics where feasible
- delete, for (k in ...), and parameter passing behave correctly for nested array references, not only top-level arrays
The latest mvn verify run for GawkCompatibilityIT shows a concentrated group of failures where Jawk still disagrees with gawk on array-versus-scalar behavior.
Observed impact:
This is now more visible because arrays-of-arrays support landed in #438, but several gawk behaviors around passing, referencing, deleting, iterating, and coercing subarrays still do not line up.
Acceptance criteria: