Sent from my Galaxy




I wish you good day.

I wish you a better day, thank you for your time, i appreciate you, sincerely not lying.

As for the orhers, if we agree that the algorithm proposed is correct, then let me remind you what I said in earlier post.

"..The fun about this proposal,  is the idea of inventing a compiler planner, that can prove at compile time what to generate.
When to unroll a loop?
When to stamp a switch function?
How to consolidate different generated functions?
...and so on..."

So if the discussion is On then I'm willing to tackle that point.

>
>
> >
> > Sent from my Galaxy
> >
> >
> > -------- Original message --------
> > From: Marcin Jaczewski <marcinjaczewski86@gmail.com>
> > Date: 8/2/24 9:40 PM (GMT+04:00)
> > To: organicoman <organicoman@yahoo.fr>
> > Cc: Marcin Jaczewski <marcinjaczewski86@gmail.com>, std-proposals@lists.isocpp.org
> > Subject: Re: [std-proposals] Function overload set type information loss
> >
> > pt., 2 sie 2024 o 18:35 organicoman <organicoman@yahoo.fr> napisał(a):
> > >
> > >
> > >
> > > Lest be charitable and assume you mean this code:
> > >
> > >
> > > ```
> > > template<typename T>
> > > void foo() { cout <<typeid(T).name()<<endl;}
> > >
> > > auto useType(effdecltype(foo)  f)
> > > {
> > >   using T = __get_ortho_at__<0>(f);
> > >   f();
> > >   return T(0);
> > > }
> > >
> > > template<typename T>
> > > void consumeType(T t)
> > > {  cout << typeid(t).name(); }
> > >
> > > int main()
> > > {
> > >     vector<effdecltype(foo)> vec;
> > >     vec.push_back(&foo<int>);
> > >     vec.push_back(&foo<double>);
> > >     vec.push_back(&foo<float>);
> > >     for(int i=0; i<5; ++i)
> > >         vec.push_back(&foo<char>);
> > >     for(const auto& F: vec)
> > >        consumeType(useType(F));
> > > }
> > > ```
> > >
> > >
> > > It can be easy fixed by helper:
> > >
> > >
> > > ```
> > > template<typename T>
> > > void consumeTypeHelper()
> > > {
> > >     foo<T>();
> > >     return consumeType(T{});
> > > }
> > >
> > > int main()
> > > {
> > >     vector<void(*)()> vec;
> > >     vec.push_back(&consumeTypeHelper<int>);
> > >     vec.push_back(&consumeTypeHelper<double>);
> > >     vec.push_back(&consumeTypeHelper<float>);
> > >     for(int i=0; i<5; ++i)
> > >         vec.push_back(&consumeTypeHelper<char>);
> > >     for(const auto& F: vec)
> > >        F();
> > > }
> > > ```
> > >
> > > Done. Problem fixed in C++11
> > >
> >
> >
> >
> >
> >
> > > Very good... nice try.
> > > Unfortunately you miss the point.
> > >
> > > I showed a loop over a set of functions returning many types.
> > > Imagine you have a container of slots, which returns different types ( i already said that many times)
> > > Given the current state of C++ you can store them only as. Let say.
> > > vector<std::any(*)(int, double)>
> > > How can you tell what type the called function returned before it was hiden inside std::any?
> > > Is it clear?
> > >
> > >
> >
> > Then show this code, I fixed your "example", now show code that can't
> > be fixed in the same way.
> > If you can't write it, then its means this code can't exist and is not
> > a real life problem.