On Mon, Nov 7, 2016 at 11:44 AM, Alex Lorenz <arpha...@gmail.com> wrote:
> arphaman added a comment.
>
> I looked at the way `HasFallthroughStmt` is used, and it didn't seem so, no.
> It's used in the following manner in AnalysisBasedWarnings.cpp:
>
> bool FallThroughDiagFull =
> !Diags.isIgnored(diag::warn_unannotated_fallthrough, D->getLocStart());
> bool FallThroughDiagPerFunction = !Diags.isIgnored(
> diag::warn_unannotated_fallthrough_per_function, D->getLocStart());
> if (FallThroughDiagFull || FallThroughDiagPerFunction ||
> fscope->HasFallthroughStmt) {
> DiagnoseSwitchLabelsFallthrough(S, AC, !FallThroughDiagFull);
> }
>
> So it seems to me that even if `HasFallthroughStmt` isn't cleared, the
> analysis won't show up anything different for a function that re-used the
> scope info, because when diagnostics are ignored and the flag actually leads
> to incorrect call to `DiagnoseSwitchLabelsFallthrough `, the analysis won't
> show any warnings since the diagnostics are ignored. And likewise if one of
> the diagnostics is enabled then it doesn't matter if `HasFallthroughStmt` is
> true, so we won't be able to observe a difference there.
Thank you for looking into this! Drat on the lack of tests, but your
analysis makes sense to me.
~Aaron
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
