Crashtests are a very low proportion of our overall automation burden. We also gain something from running a "tricky situation" in as many different configurations (opt and debug) as possible, since we may catch bugs other than the one we had in mind when writing the crashtest.
So I think we should keep doing what we do now. On Mon, Mar 30, 2015 at 9:45 PM, Xidorn Quan <quanxunz...@gmail.com> wrote: > It seems we generally add crashtests even for softening assertions to > non-crash ones. I suggest that we mark those kind of crashtests debug build > only, so that testing machines don't need to waste time on testing them on > release build where they never crash. > > I'm not sure how many crashtests are simply for assertions, hence I have no > idea how much time would it save. But I guess it might make difference if > we mark all coming assertion crashtests as debug build only. > > I discussed this with dholbert on IRC, he preferred we keep what we > currently do, since he thought marking crashtest differently according to > what is tested could increase the mental burden and add the opportunties to > make mistakes. He also thought it would be better to have wider test > coverage for less-used code paths. > > What do you think? > > - Xidorn > _______________________________________________ > dev-platform mailing list > dev-platform@lists.mozilla.org > https://lists.mozilla.org/listinfo/dev-platform > _______________________________________________ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
