Hey, The original issue reported on this bug was a pretty fundamental issue where RT just wasn't able to run for most tests, where most (all?) tests that tested the web interface had this returned:
An internal RT error has occurred. Your administrator can find more details in RT's log files. I've managed to reproduce this with the 5.0.9 packages on Unstable, and found: got warning: Wide character in subroutine entry at /<<PKGBUILDDIR>>/lib/RT/Squish.pm line 85. This is due to change in libplack-perl v1.0052, and the fix for RT was included in the 5.0.10+dfsg-1 packages via this commit: https://github.com/bestpractical/rt/commit/f8f03dd6e69dfbf4eb71e3ded0f793af4721a06d The issue with the ~50% failures with 5.0.10 appears to be due to a floating point precision mismatch in Perl and SQLite in one of the tests. I don't see it on my Intel i7, but Santiago is seeing it on their AMD EPYC. I've submitted a PR for this and will push a new 5.0.10 shortly to include the patch (and close this bug). Thanks again Santiago for your testing, and allowing me access to a VM to investigate the issue. Cheers, Andrew -- Andrew Ruthven, Wellington, New Zealand [email protected] | Catalyst Cloud: | This space intentionally left blank https://catalystcloud.nz |
