On Wed, 13 Dec 2023 21:38:43 GMT, Roger Riggs <rri...@openjdk.org> wrote:
> The test java/lang/String/CompactString/MaxSizeUTF16String.java fails when > run with -Xcomp. > > Both the java implementation and the intrinsic for StringUTF16.toBytes() > allocate memory for a copy of the string. > The java implementation of `toBytes()` throws an exception with a message in > terms of length of the string. > The intrinsic uses a generic message when allocating a byte array that is too > large for the implementation. > > Test should accept either message on the OOME exception, the message is an > implementation detail and should reflect the cause of the error and not be > confused with a general out of java heap message. The update to the definition introduces a new `@run` with `-Xcomp` so as to explicitly control the test run instead of relying on external test launch mechanisms to pass around the `-Xcomp`. That then means that the introduction of `@requires vm.flagless`, in this change, is fine. The change in the test logic itself looks reasonable to me based on what's explained in the PR description. ------------- Marked as reviewed by jpai (Reviewer). PR Review: https://git.openjdk.org/jdk/pull/17095#pullrequestreview-1781459943