Sirraide wrote: > Yeah, I'm also not opposed to this (an attribute is better than a comment > because the attribute can be analyzed)
That was my main line of reasoning here as well. > more just wondering if we want to remove the comment rather than add the > attribute. I was wondering about that too, yeah, but I’m not familiar enough w/ the surrounding code to know whether a comment/annotation is needed (or would help) here. > I think what we want to aim for is using [[fallthrough]] annotations whenever > the cases are not adjacent to one another (ignoring comments and whitespace) > and not using comments except when they're explaining why the fallthrough > happens. That is pretty much what I’m used to as well (adjacent case labels w/ nothing inbetween make it pretty clear that fallthrough was intended, after all—and we also don’t diagnose cases like that for that exact reason). So in this case, I suppose, I’ll leave it to someone more familiar w/ codegen than me to approve/reject this pr for good. https://github.com/llvm/llvm-project/pull/85921 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
