Posted on Atila's blog yesterday:

https://atilanevesoncode.wordpress.com/2016/07/18/c-is-not-magically-fast-part-2/

On Tuesday, 19 July 2016 at 02:54:37 UTC, Saurabh Das wrote:
>
> Posted on Atila's blog yesterday:
>
> https://atilanevesoncode.wordpress.com/2016/07/18/c-is-not-magically-fast-part-2/

Where is the part one ?

On Tuesday, 19 July 2016 at 03:03:38 UTC, deadalnix wrote:
> On Tuesday, 19 July 2016 at 02:54:37 UTC, Saurabh Das wrote:
>>
>> Posted on Atila's blog yesterday:
>>
>> https://atilanevesoncode.wordpress.com/2016/07/18/c-is-not-magically-fast-part-2/
>
> Where is the part one ?

https://atilanevesoncode.wordpress.com/2015/08/24/c-is-not-magically-fast/

On Tuesday, 19 July 2016 at 02:54:37 UTC, Saurabh Das wrote:
>
> Posted on Atila's blog yesterday:
>
> https://atilanevesoncode.wordpress.com/2016/07/18/c-is-not-magically-fast-part-2/

So, about D vs C++ there... last night for reasons I forget I tried replacing std::string with const char* in the C++ version, and then it got faster than D. I don't know why.

At first I thought std::string was being copied instead of being moved, but some static_asserts made me doubt that. Either way, there's no good reason I can think of for C++ to magically speed up for const char*. Hmm :(

Atila

On Tuesday, 19 July 2016 at 10:07:11 UTC, Atila Neves wrote:
> On Tuesday, 19 July 2016 at 02:54:37 UTC, Saurabh Das wrote:
>>
>> Posted on Atila's blog yesterday:
>>
>> https://atilanevesoncode.wordpress.com/2016/07/18/c-is-not-magically-fast-part-2/
>
> So, about D vs C++ there... last night for reasons I forget I tried replacing std::string with const char* in the C++ version, and then it got faster than D. I don't know why.
>
> At first I thought std::string was being copied instead of being moved, but some static_asserts made me doubt that. Either way, there's no good reason I can think of for C++ to magically speed up for const char*. Hmm :(
>
> Atila

What is the sizeof(Foo) for all of the cases?

What does "charPtr < charPtr" do in C++ compared to std::string?

Cheers, Jakob.

On Tuesday, 19 July 2016 at 03:03:38 UTC, deadalnix wrote:
> On Tuesday, 19 July 2016 at 02:54:37 UTC, Saurabh Das wrote:
>>
>> Posted on Atila's blog yesterday:
>>
>> https://atilanevesoncode.wordpress.com/2016/07/18/c-is-not-magically-fast-part-2/
>
> Where is the part one ?

https://atilanevesoncode.wordpress.com/2015/08/24/c-is-not-magically-fast/

On Tuesday, 19 July 2016 at 10:07:11 UTC, Atila Neves wrote:
> On Tuesday, 19 July 2016 at 02:54:37 UTC, Saurabh Das wrote:
>>
>> Posted on Atila's blog yesterday:
>>
>> https://atilanevesoncode.wordpress.com/2016/07/18/c-is-not-magically-fast-part-2/
>
> So, about D vs C++ there... last night for reasons I forget I tried replacing std::string with const char* in the C++ version, and then it got faster than D. I don't know why.
>
> At first I thought std::string was being copied instead of being moved, but some static_asserts made me doubt that. Either way, there's no good reason I can think of for C++ to magically speed up for const char*. Hmm :(
>
> Atila

C++ Swap may be specialized for POD.

Am 19.07.2016 um 12:07 schrieb Atila Neves:
> On Tuesday, 19 July 2016 at 02:54:37 UTC, Saurabh Das wrote:
>>
>> Posted on Atila's blog yesterday:
>>
>> https://atilanevesoncode.wordpress.com/2016/07/18/c-is-not-magically-fast-part-2/
>>
>
> So, about D vs C++ there... last night for reasons I forget I tried
> replacing std::string with const char* in the C++ version, and then it
> got faster than D. I don't know why.
>
> At first I thought std::string was being copied instead of being moved,
> but some static_asserts made me doubt that. Either way, there's no good
> reason I can think of for C++ to magically speed up for const char*. Hmm :(
>
> Atila

One thing that the D version does and the others don't is comparing UTF code points instead of bytes.

On Tuesday, 19 July 2016 at 14:39:54 UTC, Sönke Ludwig wrote:
> One thing that the D version does and the others don't is comparing UTF code points instead of bytes.

Are you sure? Autodecoding is a phobos feature, but here druntime is used for string comparison, and it usually doesn't provide autodecoding.

On Tuesday, 19 July 2016 at 15:05:30 UTC, Kagamin wrote:
> On Tuesday, 19 July 2016 at 14:39:54 UTC, Sönke Ludwig wrote:
>> One thing that the D version does and the others don't is comparing UTF code points instead of bytes.
>
> Are you sure? Autodecoding is a phobos feature, but here druntime is used for string comparison, and it usually doesn't provide autodecoding.

Yeah, auto-decoding isn't at play here.