| 2013-05-10 00:53:43 |
Paul J. Lucas |
description |
It seems that the latest integers implementation introduces a number of "wrongError" regressions, i.e., tests that used to pass, but now return the wrong error code.
With BIG_INTEGER ON, the following tests regress:
testset="prod-CastExpr.derived" testcase="cbcl-cast-unsignedByte-002"
testset="prod-CastExpr.derived" testcase="cbcl-cast-unsignedInt-002"
testset="prod-CastExpr.derived" testcase="cbcl-cast-unsignedShort-002"
testset="misc-CombinedErrorCodes" testcase="FOAR0002-5"
testset="fn-function-arity" testcase="fn-function-arity-017"
testset="fn-function-name" testcase="fn-function-name-018"
With BIG_INTEGER OFF, the following tests regress:
testset="op-numeric-equal" testcase="cbcl-numeric-equal-013"
testset="op-numeric-equal" testcase="cbcl-numeric-equal-014"
testset="op-numeric-equal" testcase="cbcl-numeric-equal-017"
testset="op-numeric-equal" testcase="cbcl-numeric-equal-018"
testset="op-numeric-mod" testcase="K2-NumericMod-5"
testset="op-numeric-mod" testcase="K2-NumericMod-7"
testset="op-numeric-mod" testcase="K2-NumericMod-8"
testset="op-numeric-multiply" testcase="cbcl-numeric-multiply-026"
testset="op-numeric-subtract" testcase="cbcl-numeric-subtract-002"
testset="op-to" testcase="rangeExpr-28"
testset="op-to" testcase="rangeExpr-29"
testset="op-to" testcase="rangeExpr-30"
testset="op-to" testcase="rangeExpr-31"
testset="op-to" testcase="rangeExpr-32"
testset="prod-CastExpr.derived" testcase="cbcl-cast-unsignedByte-002"
testset="prod-CastExpr.derived" testcase="cbcl-cast-unsignedInt-002"
testset="prod-CastExpr.derived" testcase="cbcl-cast-unsignedLong-001"
testset="prod-CastExpr.derived" testcase="cbcl-cast-unsignedShort-002"
testset="prod-CastExpr.derived" testcase="cbcl-cast-long-001"
testset="prod-CastExpr.derived" testcase="cbcl-cast-long-002"
testset="fn-avg" testcase="cbcl-avg-001"
testset="prod-Literal" testcase="K2-Literals-4"
testset="prod-Literal" testcase="K2-Literals-5"
testset="fn-function-arity" testcase="fn-function-arity-017"
testset="fn-function-name" testcase="fn-function-name-018" |
It seems that the latest integers implementation introduces a number of "wrongError" regressions, i.e., tests that used to pass, but now return the wrong error code.
With BIG_INTEGER OFF, the following tests regress:
testset="prod-CastExpr.derived" testcase="cbcl-cast-ncname-002"
testset="prod-CastExpr.derived" testcase="cbcl-cast-date-004"
testset="prod-CastExpr.derived" testcase="cbcl-cast-dateTime-004"
testset="prod-CastExpr.derived" testcase="cbcl-cast-unsignedShort-002"
testset="prod-CastExpr.derived" testcase="cbcl-cast-gYear-001"
testset="prod-CastExpr.derived" testcase="cbcl-cast-gYearMonth-001"
With BIG_INTEGER ON, the following ADDITIONAL tests regress:
testset="prod-CastExpr.derived" testcase="cbcl-cast-nonNegativeInteger-001"
testset="prod-CastExpr.derived" testcase="cbcl-cast-unsignedByte-002"
testset="prod-CastExpr.derived" testcase="cbcl-cast-unsignedInt-002"
testset="prod-CastExpr.derived" testcase="cbcl-cast-unsignedLong-002"
testset="misc-CombinedErrorCodes" testcase="FOAR0002-5"
testset="fn-function-arity" testcase="fn-function-arity-017"
testset="fn-function-name" testcase="fn-function-name-018" |
|
