On Tue, Sep 12, 2023 at 01:34:43PM -0400, Marek Polacek via Gcc-patches wrote:
> On Tue, Sep 12, 2023 at 01:27:44PM -0400, Jason Merrill via Gcc-patches wrote:
> > Tested x86_64-pc-linux-gnu, applying to trunk.
> >
> > -- 8< --
> >
> > The argument might not already be an integer.
> >
> > PR c++/111357
> >
> > gcc/cp/ChangeLog:
> >
> > * pt.cc (expand_integer_pack): Convert argument to int.
> >
> > gcc/testsuite/ChangeLog:
> >
> > * g++.dg/ext/integer-pack7.C: New test.
> > ---
> > gcc/cp/pt.cc | 2 ++
> > gcc/testsuite/g++.dg/ext/integer-pack7.C | 38 ++++++++++++++++++++++++
> > 2 files changed, 40 insertions(+)
> > create mode 100644 gcc/testsuite/g++.dg/ext/integer-pack7.C
> >
> > diff --git a/gcc/cp/pt.cc b/gcc/cp/pt.cc
> > index 838179d5fe3..b583c11eb99 100644
> > --- a/gcc/cp/pt.cc
> > +++ b/gcc/cp/pt.cc
> > @@ -3793,6 +3793,8 @@ expand_integer_pack (tree call, tree args,
> > tsubst_flags_t complain,
> > }
> > else
> > {
> > + hi = perform_implicit_conversion_flags (integer_type_node, hi,
> > complain,
> > + LOOKUP_IMPLICIT);
>
> FWIW, we have perform_implicit_conversion for this.
Is it correct to convert exactly to integer_type_node though?
Consider
#include <utility>
using std::integer_sequence;
using std::make_integer_sequence;
template<long... V>
void g(integer_sequence<long,V...>)
{}
template<typename ...T>
struct c1
{
static constexpr int value = 1;
constexpr operator int() { return value; }
constexpr operator long() { return value + 1; }
};
template<typename T>
struct R
{
using S = make_integer_sequence<long,c1<T>{}>;
R() noexcept(noexcept(g(S())))
{}
};
int main()
{
R<int>();
}
Shouldn't that invoke c1<T>{}.operator long() rather than operator int()?
I thought the conversion was supposed to be done a few lines earlier,
instead of doing
CALL_EXPR_ARG (call, 0) = hi;
do
CALL_EXPR_ARG (call, 0)
= perform_implicit_conversion_flags (TREE_TYPE (ohi), hi,
complain, LOOKUP_IMPLICIT);
or tsubst that TREE_TYPE (ohi) as well? I.e. convert to type of the
template parameter.
Jakub
