post-mortem of how the regression happened:
I did most of my testing of the fix for the demo shell on rev 1671.2.42
Jenkins was complaining of a SIGILL on the new test for the demo shell [1].
This was fixed in the test in 1671.2.46, but caused a bigger problem in the running of the demo shell, which this bug describes.
I was working off the assumption that our Jenkins CI runs the demo shell, but it actually runs mir_demo_server_basic (which isn't affected by this problem), so the problem slipped through.
My suggestion for avoiding regressing in the future is to run all of our demos (basic, shell, minimal, translucent) in the CI.
post-mortem of how the regression happened: server_ basic (which isn't affected by this problem), so the problem slipped through.
I did most of my testing of the fix for the demo shell on rev 1671.2.42
Jenkins was complaining of a SIGILL on the new test for the demo shell [1].
This was fixed in the test in 1671.2.46, but caused a bigger problem in the running of the demo shell, which this bug describes.
I was working off the assumption that our Jenkins CI runs the demo shell, but it actually runs mir_demo_
My suggestion for avoiding regressing in the future is to run all of our demos (basic, shell, minimal, translucent) in the CI.
[1] /jenkins. qa.ubuntu. com/job/ mir-mediumtests -builder- utopic- armhf/285/ console
https:/