aaron.ballman added inline comments.
================ Comment at: test/clang-tidy/modernize-use-uncaught-exceptions.cpp:64 + // CHECK-MESSAGES: [[@LINE-1]]:10: warning: 'std::uncaught_exception' is deprecated, use 'std::uncaught_exceptions' instead + // CHECK-FIXES: foo = &uncaught_exceptions; + ---------------- koldaniel wrote: > aaron.ballman wrote: > > Applying this fix will break the code so that it no longer compiles. > True, declaration of foo should be changed from type bool to int too. Since > as I can see this could cause a lot of problems, shouldn't be there only a > warning without fixits? I think the right approach here is to warn and not suggest a fix-it. ================ Comment at: test/clang-tidy/modernize-use-uncaught-exceptions.cpp:68 + // CHECK-MESSAGES: [[@LINE-1]]:22: warning: 'std::uncaught_exception' is deprecated, use 'std::uncaught_exceptions' instead + // CHECK-FIXES: res = doSomething2<uncaught_exceptions>(); + ---------------- koldaniel wrote: > aaron.ballman wrote: > > This fix seems bad. If the user accepts the fix, then the code will > > diagnose because there's no longer a matching call to `doSomething2()`. > Same type error as earlier, should a fix be applied (changing the type of the > parameter in template definition would be unlucky too, maybe a wrapper > function which could be passed to doSomething2() and does the conversion) or > only a warning? Similarly, I would only warn here as well, and not suggest a fix-it. https://reviews.llvm.org/D40787 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits