steveire added a comment. In D56444#1351247 <https://reviews.llvm.org/D56444#1351247>, @sammccall wrote:
> In D56444#1351174 <https://reviews.llvm.org/D56444#1351174>, @steveire wrote: > > > Yes - don't use RAV to traverse parents when AST-matching. > > > OK, this is certainly a much more invasive change, and isn't going to make > the release. > I understand that you don't think this crash is worth fixing for 8.0, while > others disagree. Actually, I was trying to explain what can come in the future in response to a question, but I replied with a stronger-looking opinion than I intended. I don't mind a non-invasive change like this going in now/soon, but I oppose more-drastic changes right now. > However, such nodes are very easy to reach, inside or outside of a matcher. > e.g. `callExpr(callee(functionDecl()))` will match a lambda call, and "func" > will be the `operator()`. > Having such nodes then not work with parent/ancestor matchers seems > surprising and not **obviously** the best design choice (vs e.g. visiting all > nodes including implicit, or breaking symmetry between downward and upward > traversal). I believe I have a solution to this, but this is not the place to discuss it. We can discuss it when I propose that patch :). I'll be sure to add you to the review. Repository: rC Clang CHANGES SINCE LAST ACTION https://reviews.llvm.org/D56444/new/ https://reviews.llvm.org/D56444 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits