In 29f0eb3, committed to blead yesterday (Aug 17 2026), we permitted the following test in t/op/glob.t to run "live" on non-debugging builds on Linux.
SKIP: {
skip "Debugging builds on Linux and Cygwin still problematic: GH 16869", 1
if (
($Config{osname} eq 'linux' or $Config{osname} eq 'cygwin') and
$is_debugging_build
);
fresh_perl(<<~'HERE', {});
my $glob = ("0" x 4094) . "?";
glob $glob;
HERE
is($?, 0, 'No assertion failure; GH 16869');
}
So while we still expect an assertion failure when run in a -DDEBUGGING build on Linux, we expect the test to PASS on all non-debugging builds on Linux (and elsewhere, as it happens). (We continue to handle the case for debugging builds in t/run/todo.t.)
We received one smoke-test report overnight which reports a failure in this test program in a non-debugging build on Linux. You can view that report at either of these URLs:
Of note about this report:
- Architecture is aarch64
- First FAIL in some time on this machine (host: "7of9")
- Very large number of build-time warnings in smoke-test reports from this host
- Run with these variants:
../t/op/glob.t (FAILED 19)
[stdio/perlio/locale:C.UTF-8]
[stdio/perlio/locale:C.UTF-8] -Duse64bitint
I have no access to, or experience with, the Aarch64 architecture, so I can't reproduce or diagnose it. Can someone with that access investigate? Thanks.
Related open issue: #16869
In 29f0eb3, committed to blead yesterday (Aug 17 2026), we permitted the following test in
t/op/glob.tto run "live" on non-debugging builds on Linux.So while we still expect an assertion failure when run in a
-DDEBUGGINGbuild on Linux, we expect the test to PASS on all non-debugging builds on Linux (and elsewhere, as it happens). (We continue to handle the case for debugging builds int/run/todo.t.)We received one smoke-test report overnight which reports a failure in this test program in a non-debugging build on Linux. You can view that report at either of these URLs:
https://perl.develop-help.com/db/5537374
https://perl5.test-smoke.org/report/5537374
Of note about this report:
I have no access to, or experience with, the Aarch64 architecture, so I can't reproduce or diagnose it. Can someone with that access investigate? Thanks.
Related open issue: #16869