On Mon, May 05, 2025 at 01:13:28PM -0400, Scott Talbert wrote: > On Mon, 5 May 2025, Adrian Bunk wrote: > > > Control: reopen -1 > > > > On Mon, May 05, 2025 at 04:36:04PM +0000, Debian Bug Tracking System wrote: > > > > > Date: Mon, 5 May 2025 12:23:35 -0400 (EDT) > > > From: Scott Talbert <s...@techie.net> > > > To: Chris Hofstaedtler <z...@debian.org>, 1104730-d...@bugs.debian.org > > > Subject: Re: Bug#1104730: wxwidgets3.2: FTBFS on arm* > > > > > > This seems have been some sort of transient error with the arm buildds, > > > all > > > rebuilt successfully on retry. Closing. > > > > It's not transient, the build succeeded only on the fastest arm buildds. > > > > Also looking at #1050238, is some timeout too small? > > How does the speed of amdahl compare to the rest of the arm buildds? If > it's also "fastest" then it will likely be hard to test any sort of fix > attempt for this. Other than disabling these tests. > > Looks like wxwidgets3.2 built fine on a variety of arm buildds previously. I > suspect this change in behavior is probably related to webkit2gtk changes...
The i386-only test failure is likely due to excess precision of the FPU, but most/all of the rest might just be a timeout too short. riscv64 has the same hardware on buildds and porterbox, and it is slower than all arm hardware we have. Result on the buildd: All tests passed (17060 assertions in 360 test cases) Removing both test exclusions from debian/rules: test cases: 361 | 360 passed | 1 failed assertions: 17156 | 17127 passed | 29 failed Removing both test exclusions from debian/rules plus - bool WaitEvent(int timeInMs = 1000); + bool WaitEvent(int timeInMs = 10000); All tests passed (17179 assertions in 361 test cases) > Scott cu Adrian
