From patchwork Sun Jun 27 13:10:59 2010 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Subject: [Ada] Fix bogus noreturn warning From: Steven Bosscher X-Patchwork-Id: 57095 Message-Id: To: Eric Botcazou Cc: gcc-patches@gcc.gnu.org Date: Sun, 27 Jun 2010 15:10:59 +0200 On Sun, Jun 27, 2010 at 2:45 PM, Eric Botcazou wrote: >> What is this necessary for? No front end should have to know >> *anything* about tree-flow.h. > > block_may_fallthru is declared there.  See c-typeck.c, cp/decl.c and cp/tree.c > for pre-existing counter-examples. I'm not looking for counter-examples, just stating the way things should be, not how things are (the Germans have a beautiful way of expressing this: "Soll-Stand" vs. "Ist-Stand"). Your change moves us further away from the way things should be. I hope you agree with that? Mind if I add a note, as below? Ciao! Steven Index: ada/gcc-interface/trans.c =================================================================== --- ada/gcc-interface/trans.c (revision 161470) +++ ada/gcc-interface/trans.c (working copy) @@ -33,7 +33,7 @@ #include "output.h" #include "libfuncs.h" /* For set_stack_check_libfunc. */ #include "tree-iterator.h" -#include "tree-flow.h" +#include "tree-flow.h" /* For block_may_fallthru. */ #include "gimple.h" #include "ada.h"