RFR: 8283325: US_ASCII decoder relies on String.decodeASCII being exhaustive

Roger Riggs rriggs at openjdk.java.net
Thu Mar 17 16:50:35 UTC 2022


On Thu, 17 Mar 2022 16:08:42 GMT, Claes Redestad <redestad at openjdk.org> wrote:

> Adjust `String.decodeASCII` to deal with the possibility that `StringCoding.countPositives` can return a value less than the exact number of leading positive bytes. This is the likely cause of a number of intermittent but somewhat common errors on aarch64 tests in our CI.

In the test, it might be worth a comment that the 100_000 repeats is enough to warm up the compiler and use the intrinsic.

-------------

Marked as reviewed by rriggs (Reviewer).

PR: https://git.openjdk.java.net/jdk/pull/7855


More information about the core-libs-dev mailing list