The testcase in question should only concern the parsing of UTF-8 identifiers. We added another testcase which specifically covers the regression in this case, which is the parsing of floating point values with no digits after the period (e.g. '15.' instead of '15.0').
Adding another testcase correctly points us to the issue in case we run into a regression later on. My first instinct for this issue was that it had to do with the parsing of UTF-8 identifiers, when that was not the case - so adding another testcase is necessary anyways.
Best, Arthur On 6/5/26 3:13 PM, Andreas Schwab wrote:
On Jun 05 2026, [email protected] wrote:Since the testcase only concerns the parsing of UTF-8 identifiers, work around the regression by modifying the testcase itself.No, you should not silence the error in the testsuite, because that won't fix the real error, and the failing test will remind everyone. That's the whole point of the testsuite.We opened Rust-GCC#4575 to address the issue later on, but let's restore the aarch64 build in the meantime.There is no build error, only a test failure.
OpenPGP_0x1B3465B044AD9C65.asc
Description: OpenPGP public key
OpenPGP_signature.asc
Description: OpenPGP digital signature
