On Mon, Sep 9, 2013 at 7:49 AM, Paolo Carlini <paolo.carl...@oracle.com> wrote: > Hi all, hi Gaby, > > thus a more sensible try at fixing this issue: > function.c:do_warn_unused_parameter uses "%q+D" for the error call. That > means that cp/error.c:cp_printer does: > > if (set_locus && t != NULL) > *text->locus = location_of (t); > > and it's important that location_of (t) is correct. However, for PARM_DECLs, > location_of (still) does: > > > if (TREE_CODE (t) == PARM_DECL && DECL_CONTEXT (t)) > t = DECL_CONTEXT (t); > else if ... > ... > > which means that for those we end up with the location of a FUNCTION_DECL or > anyway not exactly the location of the PARM_DECL itself. I went through the > C++ front-end and currently there are very few cases if any where > location_of may be fed a PARM_DECL, and the very few seem Ok to me. The > testsuite passes on x86_64-linux of course, thus I'm proposing the below. > IMHO now that we are in Stage 1 the risk of breaking locations somewhere > else is low enough. By the way, as discussed today elsewhere, the C > front-end is already Ok.
Patch OK; thanks! -- Gaby
