On 3/7/22 14:41, Patrick Palka wrote:
instantiate_non_dependent_expr_sfinae instantiates only potentially constant expressions
Hmm, that now strikes me as a problematic interface, as we don't know whether what we get back is template or non-template trees.
Maybe we want to change instantiate_non_dependent_expr to checking_assert that the argument is non-dependent (callers are already checking that), and drop the potentially-constant test?
but when processing a non-dependent decltype operand we want to instantiate it even if it's non-constant since such non-dependent decltype is always resolved ahead of time. Currently finish_decltype_type uses the former function, which causes us to miss issuing a narrowing diagnostic for S{id(v)} in the below testcase because we never instantiate this non-constant non-dependent decltype operand. So this patch makes finish_decltype_type use i_n_d_e_internal instead of _sfinae. And afterward, we need to keep processing_template_decl cleared for sake of the later call to lvalue_kind, which handles templated and non-templated COND_EXPR differently. Otherwise we'd incorrectly reject the declaration of g in cpp0x/cond2.C with: error: ‘g’ declared as function returning a function Bootstrapped and regtested on x86_64-pc-linux-gnu, does this look OK for trunk? PR c++/104823 gcc/cp/ChangeLog: * semantics.cc (finish_decltype_type): Use i_n_d_e_internal instead of _sfinae when instantiating a non-dependent decltype operand, and keep processing_template_decl cleared afterward. gcc/testsuite/ChangeLog: * g++.dg/cpp0x/Wnarrowing19.C: New test. --- gcc/cp/semantics.cc | 11 ++++++++++- gcc/testsuite/g++.dg/cpp0x/Wnarrowing19.C | 8 ++++++++ 2 files changed, 18 insertions(+), 1 deletion(-) create mode 100644 gcc/testsuite/g++.dg/cpp0x/Wnarrowing19.C diff --git a/gcc/cp/semantics.cc b/gcc/cp/semantics.cc index 07cae993efe..66d90c2f7be 100644 --- a/gcc/cp/semantics.cc +++ b/gcc/cp/semantics.cc @@ -11217,6 +11217,8 @@ finish_decltype_type (tree expr, bool id_expression_or_member_access_p, /* decltype is an unevaluated context. */ cp_unevaluated u;+ processing_template_decl_sentinel ptds (/*reset=*/false);+ /* Depending on the resolution of DR 1172, we may later need to distinguish instantiation-dependent but not type-dependent expressions so that, say, A<decltype(sizeof(T))>::U doesn't require 'typename'. */ @@ -11232,7 +11234,14 @@ finish_decltype_type (tree expr, bool id_expression_or_member_access_p, } else if (processing_template_decl) { - expr = instantiate_non_dependent_expr_sfinae (expr, complain); + /* Instantiate the non-dependent operand to diagnose any ill-formed + expressions. And keep processing_template_decl cleared for the rest + of the function (for sake of the call to lvalue_kind below, which + handles templated and non-templated COND_EXPR differently). */ + processing_template_decl = 0; + /* Since we want to do this even for non-constant expressions, we + use i_n_d_e_internal here instead of _sfinae. */ + expr = instantiate_non_dependent_expr_internal (expr, complain); if (expr == error_mark_node) return error_mark_node; } diff --git a/gcc/testsuite/g++.dg/cpp0x/Wnarrowing19.C b/gcc/testsuite/g++.dg/cpp0x/Wnarrowing19.C new file mode 100644 index 00000000000..bd9fd2eb6f9 --- /dev/null +++ b/gcc/testsuite/g++.dg/cpp0x/Wnarrowing19.C @@ -0,0 +1,8 @@ +// PR c++/104823 +// { dg-do compile { target c++11 } } + +struct S { S(int); }; + +double id(double v); + +template<class> auto f(double v) -> decltype(S{id(v)}); // { dg-error "narrowing" }
