JonasToth added a comment. Your point is valid, that the decision of what to analyze should be done outside. My const-correctness check does analyze all versions of the templated function, because it just matches on `functionDecl(compoundStmt())`.
Maybe we just need some experience with real world code. The const-correctness thing is close to finish in its first version. Then we can exercise its results. Am 11.09.2018 um 18:15 schrieb Shuai Wang via Phabricator: > I see, it's the conflicting results you're going after :) > Good news is that we actually don't analyze all versions, we only analyze > the version (instantiated or not) corresponding to the "scope" stmt passed > into the constructor. Semantic-wise I feel this makes sense because if we're > given an instantiated version we shouldn't bail out because nothing is > type-dependent anymore in the instantiated version. > Also I think conflicts won't happen much in practice, most (all?) checks > naturally pass in the uninstantiated version, in order to pass in an > instantiated version a check needs to: > > - Find an instantiation point > - Match and extract the function decl from the callExpr > - Extract function body compontStmt from the function decl at that point the > check owner likely knows pretty well what they're doing and shouldn't be > surprised that analyze results conflicts if they happen to also analyze an > uninstantiated version. > > Repository: > > rCTE Clang Tools Extra > > https://reviews.llvm.org/D50883 Repository: rCTE Clang Tools Extra https://reviews.llvm.org/D50883 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
