I'm pretty sure NullStmt should still never be in the CFG, even within an AttributedStmt. For the AttributedStmt itself, please add it to the "Cases we intentionally don't evaluate" section. That allows checkers to do pre/post-stmt checks, but doesn't generate a new node if there's no actual change in state. We can always change this later if we ever want Core handling any statement attributes.
http://llvm-reviews.chandlerc.com/D1030 BRANCH master ARCANIST PROJECT clang _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
