On Fri, 28 Sep 2018 at 10:23, Ben Coman <[email protected]> wrote:
> Just a random idea... > when running Testrunner locally and a random test causes the image to > freeze, its a bit of an effort to identify which test was being run. You > need to restart and then start bisecting the tests and perhaps it is > intermittent so the bisect doesn't work... > > but now that PharoLauncher has been mainstreamed > maybe it could provide a centralised Reporting Centre > i.e. open a port for launched images to report the test about-to-commence > as well as the run time. Then it would be simple to see which test froze > the system. Also it could keep a record individual test runtimes, to > alleviate fears when the system is not frozen but just a long running test. > > Extending the idea, while CI testing is very useful, we get these sporadic > errors that are hard to replicate locally. A local PharoLauncher driven > system for repetitively launching an image auto-running tests might be good > to gather statistics on sporadic failures to then drill down on the > troublemakers. > > Plus with the goal of having an object based logging system, a viewer is > needed and perhaps PharoLauncher would be a logical place to provide a > logging service to gather and especially to view log-objects reporting by > launched images. > side thought, such a system might be good for watching test progress on headless IoT devices. > > of course, resources required to do it, > but just thought I share the idea. > > cheers -ben >
