Comments on: Templates as first-class citizens in C++11

By: qicosmos

qicosmos — Tue, 26 Jul 2016 03:34:00 +0000

call_impl can be resolved by std::index_sequence, here is the code:
template
auto call_impl(Arg&& arg, const std::index_sequence&) const ->decltype(std::get(functions_)(std::forward(arg)))
{
return std::get(functions_)(std::forward(arg));
}

template
auto call_impl(Arg&& arg, const std::index_sequence&) const ->decltype(call_impl(std::get(functions_)(std::forward(arg)), std::index_sequence{}))
{
return call_impl(std::get(functions_)(std::forward(arg)), std::index_sequence{});
}

template
auto call(Arg&& arg) const-> decltype(call_impl(std::forward(arg), std::make_index_sequence{}))
{
return call_impl(std::forward(arg), std::make_index_sequence{});
}

By: qicosmos

qicosmos — Tue, 12 Jul 2016 07:33:00 +0000

Hi, Victor, can you improve the code with c++14, now vs2015 and gcc6.0 are surpport most c++11 features. I think c++14 will make the code clean and simple.

By: Victor Laskin

Victor Laskin — Tue, 12 Jan 2016 07:28:00 +0000

In reply to Goobley. Thanks. No problem with that - you could use code as you like. As for the video - it will be available soon (I expect it to be at jan or feb).

By: Goobley

Goobley — Mon, 11 Jan 2016 23:21:00 +0000

Hey, great job with this, I’ve based some code on some of this, in the process of becoming a small header-only library. Any chance you could officially state it being under a certain license so I can properly credit it. (I’d rather anything non-GPL). I found your talk via your MeetingCPP 2015 slides, any chance there’ll be a video of the talk that goes with them? Cheers!

By: Victor Laskin

Victor Laskin — Wed, 29 Apr 2015 16:08:00 +0000

In reply to qicosmos. Hello, yes, it's less natural - but you can switch it with some other syntax you prefere. The main idea was to show that such decomposition is possible.

By: qicosmos

qicosmos — Wed, 29 Apr 2015 03:30:00 +0000

Hi Victor, i don’t think “<<" is a good choice, it's not natural, your old post "functional piping" is better.

By: Victor Laskin

Victor Laskin — Sun, 08 Mar 2015 10:35:00 +0000

In reply to Dan George. Yes, thanks. Sure, it's obvious typo during manual macro expansion. ps. I can't wait for C++14 where result types could be written as decltype(auto) and whole code will be a lot shorter.

By: Dan George

Dan George — Sun, 08 Mar 2015 04:46:00 +0000

Hi Victor, thanks for another inspiring post. I think there is a typo in the tfn_print wrapper: dectype(X( s/b dectype(print(, right?

By: Victor Laskin

Victor Laskin — Sat, 07 Mar 2015 21:49:00 +0000

In reply to AntonBikineev. Anton, Thanks. No, just && was missing there. I fixed it.

By: AntonBikineev

AntonBikineev — Sat, 07 Mar 2015 20:47:00 +0000

I just read your first code snippet and got a question. Is there any sense to use std forward for non universal references?

By: Victor Laskin

Victor Laskin — Thu, 05 Mar 2015 18:36:00 +0000

In reply to ashenwolf. Hello, thanks for comment. No, it's not just academic - you can easily compile gist from link at the end of post. It may be more problematic to get your developers think in more functional way, but starting to use the technic is easy. As this is quite new approach I can't provide any information about long term maintainability (A lot of people still can't use even c++11 in production).