https://www.quora.com/Why-hasnt-D-started-to-replace-C++

Andrei

On Tue, Jan 30, 2018 at 03:45:44PM -0500, Andrei Alexandrescu via Digitalmars-d wrote:
> https://www.quora.com/Why-hasnt-D-started-to-replace-C++
[...]

I actually agree with all of his points, except one: C++'s "enormous performance advantage"?!  Is he being serious?  Or is his view biased by dmd's, erm, shall we say, "suboptimal" optimizer?

Just yesterday I was experimenting with ldc while reviewing a Phobos PR, and it showed that LDC was well able to reduce iteration over a range down to a native asm loop of the same order of efficiency as an equivalent C++ loop.  So, the cost of D (arguably better) abstractions simply *isn't there*.  Both languages compile to the *same* native code. Where's C++'s "enormous performance advantage?"  I'm not seeing it, except in this article, and, presumably, in the author's imagination.

And of course, in the comments section there's the obligatory reference to the Tango/Phobos split in D1.  Eye-roll.  That one never gets old, it seems, even though it's already ancient history that the modern D community has mostly forgotten about.  The internet is good for disseminating information, and also great for prolonging ignorance.

Enough of this nonsense.  Let me get back to coding... in D. :-D


T

-- 
What is Matter, what is Mind? Never Mind, it doesn't Matter.

On Tuesday, 30 January 2018 at 21:02:13 UTC, H. S. Teoh wrote:
> On Tue, Jan 30, 2018 at 03:45:44PM -0500, Andrei Alexandrescu via Digitalmars-d wrote:
>> https://www.quora.com/Why-hasnt-D-started-to-replace-C++
> [...]
>
> I actually agree with all of his points, except one: C++'s "enormous performance advantage"?!  Is he being serious?  Or is his view biased by dmd's, erm, shall we say, "suboptimal" optimizer?

Well, I understand what you mean, but there are more performance related options for C++ than for any other language at the moment. I don't use them, most people don't, but they exist. So for people that care a lot about throughput there basically is no other "complete" alternative...

On Tue, Jan 30, 2018 at 09:26:30PM +0000, Ola Fosheim Grøstad via Digitalmars-d wrote:
> On Tuesday, 30 January 2018 at 21:02:13 UTC, H. S. Teoh wrote:
> > On Tue, Jan 30, 2018 at 03:45:44PM -0500, Andrei Alexandrescu via Digitalmars-d wrote:
> > > https://www.quora.com/Why-hasnt-D-started-to-replace-C++
> > [...]
> > 
> > I actually agree with all of his points, except one: C++'s "enormous performance advantage"?!  Is he being serious?  Or is his view biased by dmd's, erm, shall we say, "suboptimal" optimizer?
> 
> Well, I understand what you mean, but there are more performance related options for C++ than for any other language at the moment. I don't use them, most people don't, but they exist. So for people that care a lot about throughput there basically is no other "complete" alternative...

Specific examples, please. What are some examples of these "performance related options"?


T

-- 
If it's green, it's biology, If it stinks, it's chemistry, If it has numbers it's math, If it doesn't work, it's technology.

Thats not completly true, last time I tried some of best c++ version from http://benchmarksgame.alioth.debian.org/. I was able to write much idiomatic D code which was faster than c++ witch use some specific libraries and so on. So my experience is that D is same or faster than C++

On Tue, Jan 30, 2018 at 10:26 PM, Ola Fosheim Grøstad via Digitalmars-d < digitalmars-d@puremagic.com> wrote:

> On Tuesday, 30 January 2018 at 21:02:13 UTC, H. S. Teoh wrote:
>
>> On Tue, Jan 30, 2018 at 03:45:44PM -0500, Andrei Alexandrescu via Digitalmars-d wrote:
>>
>>> https://www.quora.com/Why-hasnt-D-started-to-replace-C++
>>>
>> [...]
>>
>> I actually agree with all of his points, except one: C++'s "enormous performance advantage"?!  Is he being serious?  Or is his view biased by dmd's, erm, shall we say, "suboptimal" optimizer?
>>
>
> Well, I understand what you mean, but there are more performance related options for C++ than for any other language at the moment. I don't use them, most people don't, but they exist. So for people that care a lot about throughput there basically is no other "complete" alternative...
>
>

On Tuesday, 30 January 2018 at 21:30:06 UTC, H. S. Teoh wrote:
> On Tue, Jan 30, 2018 at 09:26:30PM +0000, Ola Fosheim Grøstad via Digitalmars-d wrote:
> Specific examples, please. What are some examples of these "performance related options"?

Supported concurrency options and tuning etc.

On Tue, Jan 30, 2018 at 09:49:46PM +0000, Ola Fosheim Grøstad via Digitalmars-d wrote:
> On Tuesday, 30 January 2018 at 21:43:45 UTC, Daniel Kozak wrote:
> > Thats not completly true, last time I tried some of best c++ version from http://benchmarksgame.alioth.debian.org/. I was able to write much idiomatic D code which was faster than c++ witch use some specific libraries and so on. So my experience is that D is same or faster than C++
> 
> I'm not thinking about those. I'm thinking about supported tooling that makes you more productive when writing code with high throughput.

That's not what the article says. The article says, and I quote:

	C++’s enormous performance advantage (extremely efficient native
	code)

Meaning, the "enormous performance advantage" is because of "extremely eefficient native code".  I don't argue that C++ has extremely efficient native code. But so has D. So the claim that C++ has an "enormous performance advantage" over D is specious.


T

-- 
Nearly all men can stand adversity, but if you want to test a man's character, give him power. -- Abraham Lincoln

On Tuesday, 30 January 2018 at 21:43:45 UTC, Daniel Kozak wrote:
> Thats not completly true, last time I tried some of best c++ version from http://benchmarksgame.alioth.debian.org/. I was able to write much idiomatic D code which was faster than c++ witch use some specific libraries and so on. So my experience is that D is same or faster than C++

I'm not thinking about those. I'm thinking about supported tooling that makes you more productive when writing code with high throughput.

Yes, than you are right

On Tue, Jan 30, 2018 at 10:49 PM, Ola Fosheim Grøstad via Digitalmars-d < digitalmars-d@puremagic.com> wrote:

> On Tuesday, 30 January 2018 at 21:43:45 UTC, Daniel Kozak wrote:
>
>> Thats not completly true, last time I tried some of best c++ version from http://benchmarksgame.alioth.debian.org/. I was able to write much idiomatic D code which was faster than c++ witch use some specific libraries and so on. So my experience is that D is same or faster than C++
>>
>
> I'm not thinking about those. I'm thinking about supported tooling that makes you more productive when writing code with high throughput.
>
>

On Tuesday, 30 January 2018 at 21:49:39 UTC, H. S. Teoh wrote:
> Meaning, the "enormous performance advantage" is because of "extremely eefficient native code".  I don't argue that C++ has extremely efficient native code. But so has D. So the claim that C++ has an "enormous performance advantage" over D is specious.

Well, it isn't relevant for those people who would adopt D anyway.

Of course, C++ and Java have some advantages by being so large that there is a market for commercial specialty solutions and services... Although most C++ and Java programmers use tooling that is essentially free (well except perhaps the IDE), so for most of them it won't matter.

Even when smaller languages try to implement such tooling/features there isn't a large enough user base to harness the implementation (since even for big languages the actual user base for those features are low), so it is very hard for smaller languages to branch into those special niches unless the whole language feature set is geared towards a specific niche... but that harms adoption too...