tigerleapgorge added a comment. In https://reviews.llvm.org/D30430#688146, @rjmccall wrote:
> The C++98 behavior here is not really vital to test precisely; it's just > minor differences in what gets instantiated and when. Hi John, my main concern with CHECK-NOT appearing between CHECK lines is that it makes the test brittle to the order of the IR. For this reason I have separated out the CHECK-NOTs to go under a seperate prefix with a separate RUN line. If you would like to old way of a single RUN line, I can do that too. > I think it's fine to just update the run line to -std=c++11 for things like > this. Done. I have deleted C++98 RUN lines and default (no -std) RUN lines. > But if you really want to test both configurations, this LGTM, although > please leave a comment in the test explaining that it's just trying to > account for differences in instantiation order between C++98 and C++11. I have added 2 lines of comments. One explains why CHECK-DAG is used instead of CHECK. Another explains why CHECK2-NOT is used instead of CHECK-NOT https://reviews.llvm.org/D30430 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
