The problem that took him 5 years to fix in C++, I solved in a minute with D

[Paulo Pinto]

On Thursday, 10 March 2022 at 02:01:01 UTC, matheus wrote:
> Hi,
>
> Today someone sent me this video: 
> https://www.youtube.com/embed/ABg4_EV5L3w
> (C++ Weekly - Ep 313 - The `constexpr` Problem That Took Me 5 
> Years To Fix!).
>
> It's a 26 minute long video of a guy trying to generate a 
> concatenated string during CT in C++.
>
> I was flabbergasted by the problems this poor guy had in front 
> of him for a thing that shouldn't seem so complex (In my 
> opinion of course). And more yet that he has being struggling 
> with this for 5 years.
>
> So I decided to try the same with D, it was 14 lines[1] of code 
> written in less than a minute and it just worked on the first 
> try. Then I looked the Assembly code:
>
> .L.str:
>         .asciz  "hello world, hello world, hello world, "
>
> After that I even asked over the D IRC channel if it was just 
> this? And a user confirmed that my string was being generated 
> through CT as the Assembly indicated.
>
> Again incredible. My first programming language was C, then in 
> 2000-ish I had my first contact with C++ and I just hated and 
> went back to C and then while looking for alternatives I 
> stumbled on D which I use till this day as a "super C".
>
> But now after so many years I couldn't believe the C++ is still 
> struggling even to this day with this type of thing.
>
> I'd like to quote this comment from the video:
>
>> meowsqueak:
>> "Well done solving this puzzle, but I’m left in shock. The 
>> need to employ a template, static constexpr variables, a 
>> lambda function, and four utility functions just to 
>> compile-time generate a static string in C++ is an 
>> embarrassment."[2]
>
> And indeed it is an embarrassment. Well D may have it flaws but 
> I always found it lot time better or less restrictive language 
> compared to others, especially C++.
>
> Finally I'd like to take this moment and thank for all the 
> effort.
>
> Matheus.
>
> [1] - Here is the snippet I wrote: 
> https://godbolt.org/z/MsT36Prx4
> [2] - Youtube comment: 
> https://www.youtube.com/watch?v=ABg4_EV5L3w&lc=Ugzb7kTLS8GNS9bK07F4AaABAg
>
> PS: I'm a ESL, sorry for any English mistakes.

Well, a language alone doesn't make ecosystems, D is a good 
language, but after 10 years doesn't hold a candle to CUDA, 
Metal, Unreal, SYSCL, AUTOSAR/MISRA certifications, 
iOS/Android/Windows SDK, High Energy Physics research, 
PyTorch/Tensorflow, LLVM/GCC, ....

Before bashing C++'s design, which does indeed have plenty of 
flaws with several books describing them, what about finishing 
all those previews that are already 5 years and longer in the 
making and finally decide what domain D is good at?

The way I see it, C# 10 and C++ 20 have mostly catched up in 
where D was 10 years ago, Rust and Go have found out a way to 
push them into devs regardless of what they think about them, 
thanks all those Cloud Native Foundation projects, Khronos is 
slowly adopting Rust alongside C++ on some of their ongoing 
standard discussions.

Creating such ecosystems is what D should worry about, not how 
great a bare bones compiler is against the competition.