On Sat, Jan 26, 2019 at 07:22:17PM -0500, Jason Merrill wrote:
> On Fri, Jan 25, 2019 at 4:20 PM Marek Polacek <pola...@redhat.com> wrote:
> > On Fri, Jan 25, 2019 at 10:05:00AM -0500, Jason Merrill wrote:
> > > On 1/24/19 7:17 PM, Marek Polacek wrote:
> > > > On Wed, Jan 23, 2019 at 03:34:04PM -0500, Jason Merrill wrote:
> > > > > On Wed, Jan 23, 2019 at 12:57 PM Marek Polacek <pola...@redhat.com>
> > > > > wrote:
> > > > > >
> > > > > > On Wed, Jan 23, 2019 at 09:00:36AM -0500, Jason Merrill wrote:
> > > > > > > I was talking about digest_init, not reshape_init. digest_init
> > > > > > > calls
> > > > > > > convert_for_initialization.
> > > > > >
> > > > > > /facepalm
> > > > > >
> > > > > > So yes, digest_init calls convert_for_initialization which will end
> > > > > > up
> > > > > > calling perform_implicit_conversion_flags which could call
> > > > > > convert_like_real
> > > > > > where the narrowing warnings are given, but it doesn't, we go to
> > > > > > this case:
> > > > > >
> > > > > > else if (processing_template_decl && conv->kind != ck_identity)
> > > > > > {
> > > > > > /* In a template, we are only concerned about determining the
> > > > > > type of non-dependent expressions, so we do not have to
> > > > > > perform the actual conversion. But for initializers, we
> > > > > > need to be able to perform it at instantiation
> > > > > > (or instantiate_non_dependent_expr) time. */
> > > > > > expr = build1 (IMPLICIT_CONV_EXPR, type, expr);
> > > > > >
> > > > > > finish_decltype_type throws away the expression because it's not
> > > > > > dependent, and
> > > > > > only uses its type. So narrowing remains undetected. Not sure if
> > > > > > I should mess
> > > > > > with perform_implicit_conversion_flags.
> > > > >
> > > > > Let's try that; this is a situation where the comment is incorrect.
> > > > > Perhaps just call check_narrowing here if appropriate, rather than go
> > > > > through the whole conversion machinery.
> > > >
> > > > I have not been successful.
> > > >
> > > > First, I modified perform_implicit_conversion_flags to go the
> > > > convert_like
> > > > route when dealing with something non-dependent. That breaks e.g. in
> > > > build_value_init:
> > > > 346 /* The AGGR_INIT_EXPR tweaking below breaks in templates. */
> > > > 347 gcc_assert (!processing_template_decl
> > > > 348 || (SCALAR_TYPE_P (type) || TREE_CODE (type) ==
> > > > ARRAY_TYPE));
> > > > Even if I restrict the convert_like way for non-dependent exprs in a
> > > > template
> > > > for scalars, it still breaks elsewhere, e.g. constexpr-template3.C
> > > > where it
> > > > complains about taking the address of an rvalue.
> > > >
> > > > Second, I added check_narrowing to the processing_template_decl case in
> > > > perform_implicit_conversion_flags. That works except it breaks
> > > > constexpr-inst1.C -- we no longer get the error. That's because
> > > > currently
> > > > check_narrowing in finish_compound_literal calls maybe_constant_init,
> > > > which
> > > > calls instantiate_constexpr_fns and we get the desired diagnostic. But
> > > > if
> > > > I move check_narrowing to perform_implicit_conversion_flags, we no
> > > > longer
> > > > call it in this case -- processing_template_decl is 0 so we call
> > > > convert_like
> > > > but that doesn't do the trick.
> > > >
> > > > So, back to the patch that leaves check_narrowing in
> > > > finish_compound_literal?
> > >
> > > That patch still needs a test for the aggregate case.
> >
> > Ok, this is a version with Wnarrowing16.C added.
> >
> > ...but we still don't warn for the TYPE_NON_AGGREGATE_CLASS case in
> > finish_compound_literal, so the nightmare continues.
>
> Alas. Are you going to keep looking at that, or would you like me to take
> over?
I think I should keep poking at it, but I welcome any advice. But perhaps we
should leave this for GCC 10?
I'll add the new test to the PR.
> > - if (SCALAR_TYPE_P (type)
> > - && !BRACE_ENCLOSED_INITIALIZER_P (compound_literal)
> > + if (!BRACE_ENCLOSED_INITIALIZER_P (compound_literal)
> > && !check_narrowing (type, compound_literal, complain))
> > return error_mark_node;
>
> Does this hunk actually make a difference? It looks like
> check_narrowing only does anything for arithmetic types.
No difference (I ran the testsuite with some additional checking).
> OK with or without this hunk.
Thanks for the reviews.
Marek
