Also sprach Gregory Szorc: > On Tue, Jun 26, 2018 at 3:45 PM, Dave Townsend <dtowns...@mozilla.com> > wrote: > >> This doesn't seem to answer the question. I frequently do patches >> where I can't reliably guess the extent of test breakage and want >> to run a full set to be on the safe side. What try configuration >> gives us tier 1 results only? > > I agree that the default behavior of `-p all -u all -t all` is > running too much and confusing people. I filed bug 1471404 to change > the behavior. > > Will that be sufficient to fix the issue? > > FWIW there were also some discussions at the all hands about in-tree > profiles for `mach try` to make it easier for people to perform > common test scenarios. Would this be a better solution?
For as long as we can’t reliably track which changed files impact what tests, pre-programmed test scenarios seem like a good idea. I’ve often found myself sharing my try syntax strings with new contributors. I have these exposed as environment variables for easy access, e.g. "./mach try $trymn”. If I instead could tell them to run a given test scenario which included a recipe of most likely offenders that would be a big improvement compared to try-fuzzy or copying—what to a new contributor might seem like black magic—a string. _______________________________________________ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
