On Friday 2014-11-21 16:32 +0000, Gijs Kruitbosch wrote: > On 20/11/2014 17:14, L. David Baron wrote: > >>On 20/11/14 17:56, Boris Zbarsky wrote: > >>>Ah, we can't. We can whitelist the number of assertions in a mochitest > >>>(or a number range if the number is not quite stable), but not the text > >>>of the assertion. > > > >On Thursday 2014-11-20 18:05 +0100, David Rajchenbach-Teller wrote: > >>I believe that we can provide something less fragile than that. > > > >It's not all that fragile, since most tests don't have known > >assertions, so in that vast majority of tests, we have test coverage > >of regressions of assertion counts. > > Sure, although there are also a small number of tests which define > assertion counts of "between 10 and 18" or something like this... > and mostly when things change there (especially if in the positive > direction) we update those counts and move on with life. It would be > better if we had a more detailed understanding of why those > assertions are fired and which they are.
In general, we should have bugs filed on the assertions, and comments next to the assertion counts pointing to those bugs, just like we do for known failures. It wasn't practical to do that when switching on the assertion checking initially, though, just given the difficulty of getting the checking landed at all while the state of the tree was changing. (I believe I still have the data used for that, though, so I could probably go back and file the ones that are still present.) -David -- 𝄞 L. David Baron http://dbaron.org/ 𝄂 𝄢 Mozilla https://www.mozilla.org/ 𝄂 Before I built a wall I'd ask to know What I was walling in or walling out, And to whom I was like to give offense. - Robert Frost, Mending Wall (1914)
signature.asc
Description: Digital signature
_______________________________________________ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
