srhines added a comment. In https://reviews.llvm.org/D36728#842644, @lhames wrote:
> The preferred solution to this is actually to wrap the call with cantFail > (See > http://llvm.org/docs/ProgrammersManual.html#using-cantfail-to-simplify-safe-callsites) > -- it will handle both the assertion and consumption of the value for you, > and will simplify calls that return an Expected<T>. Is it ok to drop the assertion in that case (and convert it to a comment)? I didn't want to alter too much of this check, since perhaps the original author(s) were more skeptical about this breaking (hence the assertion). Something like: // Replacements must not conflict since ranges have been merged. llvm::cantFail(FakeReplaces.add(...)); Repository: rL LLVM https://reviews.llvm.org/D36728 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
