On Tuesday, 28 November 2017 at 08:33:20 UTC, Mike Parker wrote:
> On Tuesday, 28 November 2017 at 04:35:04 UTC, IM wrote:
>> [...]
>
> This is something that has gone in fits and starts over the years because of a lack of dedicated manpower, but the pace has been gradually to picking up. As of recently, I'm working on several tasks in this direction, big and small, with the support of the D Foundation. For example, we now have a D Language Foundation channel on youtube [1] where I'm currently in the process of collecting DConf videos that are scattered around different sites and accounts (working on 2014 first, since several of the video links on that edition of dconf.org were broken). It's not ready for announcement yet, but I hope to be there by the end of the year. There are a number of other things I'm looking at that have tended to slip through the cracks because they've been overlooked or no one has stepped in to do them.
>
> [...]

Since Mike started the official D blog last summer, downloads of the reference compiler are up 90%:

http://erdani.com/d/downloads.daily.png

I don't think that's a coincidence and attribute a significant chunk of that to his efforts and those who wrote posts, which is why I suggested starting an official blog years ago.

On 11/27/2017 10:24 PM, Dmitry Olshansky wrote:
> On Tuesday, 28 November 2017 at 06:12:19 UTC, Walter Bright wrote:
>> On 11/27/2017 9:11 PM, rikki cattermole wrote:
>>> On 28/11/2017 5:03 AM, Walter Bright wrote:
>>>> On 11/27/2017 6:55 PM, John wrote:
>>>>> Should add optlink to that list, would love to see it converted to D!
>>>>
>>> We have discussed this on Discord a little bit lately.
>>> What we are hoping for is a D dmc+libc updated to use dmd-be. Potentially allowing us to use LLVM's linker but with dmc's libc as well. Giving us out of the box experience for 64bit.
>>>
>>> It would be nice, but well, your site would need a lot of changes to go in this direction.
>>
>> Yes, I've thought about making dmc++ 64 bit, but there'd be a fair amount of work (mostly upgrading SNN to 64 bits.)
> 
> We could also convert that libc to D ;)
> Seriously betterC mode would make that way easier and more fun, is it on GitHub?

Yes, and I should finish boost licensing it!

It's written in old-fashioned C code, and a fair bit of assembler. Every line of it would have to be reviewed for 64 bit portability, and there's no test suite :-(

The good news is it has been pretty darned reliable.

There's also the STL library, which is pretty complex.


> Actually Herb Sutter shared once that Microsoft used C++ (as in templates C++) to reimplement a significant chunk of its libc with great success. Less code, less ifdef hell and macro abuse I think were presented as advantages.

Yes, I came late to the game of not using ifdef hell. I'm pretty proud of the near complete absence of version() statements in the dmd front end. It didn't start out that way!

On Tuesday, 28 November 2017 at 06:58:58 UTC, Elronnd wrote:
> In that case, why is libstdc++ 12MB, while libphobos2 is half the size, at 5.5MB?

I havent checked, if true then probably because it contains code that goes beyond the minimal requirements (legacy, bloat, portability, tuning, etc). Phobos contain more application oriented APIs than C++17.

On 11/27/2017 11:26 PM, Paulo Pinto wrote:
> Yes, the new MSVCRT.dll, is implemented in C++.
> 
> https://blogs.msdn.microsoft.com/vcblog/2014/06/10/the-great-c-runtime-crt-refactoring/ 
> 
> 
> After Midori and Longhorn's failure, there has been a migration effort to slowly get rid of C and focus on C++ for lower level stuff and .NET Native for everything else, at least on what concerns kernel, desktop and UWP.

My experience using BetterC for this task should produce much better results than C++!

On Tuesday, 28 November 2017 at 06:12:19 UTC, Walter Bright wrote:
>
> Yes, I've thought about making dmc++ 64 bit, but there'd be a fair amount of work (mostly upgrading SNN to 64 bits.)

Could I help with that?

I'm familiar with x86 assembly, including "mixed" one that use the same source for 32-bit and 64-bit.

I'd say porting 32-bit assembly to 64-bit assembly in x86 is way faster/fool-proof than removing that assembly.

On 11/28/2017 2:57 AM, Guillaume Piolat wrote:
> On Tuesday, 28 November 2017 at 06:12:19 UTC, Walter Bright wrote:
>>
>> Yes, I've thought about making dmc++ 64 bit, but there'd be a fair amount of work (mostly upgrading SNN to 64 bits.)
> 
> Could I help with that?
> 
> I'm familiar with x86 assembly, including "mixed" one that use the same source for 32-bit and 64-bit.
> 
> I'd say porting 32-bit assembly to 64-bit assembly in x86 is way faster/fool-proof than removing that assembly.

Yes, you can. I appreciate the offer! I'll get back to you.

On 2017-11-28 00:13, Walter Bright wrote:

> I suppose I should write one :-) It was a very satisfying project. I'm looking at converting all my C projects still in use (like 'make') to D. BetterC has removed the last barriers to it.

Why would druntime be a barrier for you for those projects?

-- 
/Jacob Carlborg

On Tuesday, 28 November 2017 at 05:18:42 UTC, Adam D. Ruppe wrote:
> On Tuesday, 28 November 2017 at 04:52:52 UTC, bauss wrote:
>> You're not measuring what you think for the Java program. Did you calculate the runtime and JIT initialization time and subtracted that from the actual execution time? Otherwise your benchmark isn't sufficient.
>
> For small programs, startup time is fair to consider since the end user still has to deal with that too.
>
> But for larger programs, I suspect it would disappear too.

Not when the startup time is in milliseconds. If it was a large program taking minutes to startup, but that's not the case at all. A user will barely even be able to tell the difference between 50ms.

On Tuesday, 28 November 2017 at 08:58:46 UTC, Joakim wrote:
>
> Since Mike started the official D blog last summer, downloads of the reference compiler are up 90%:
>
> http://erdani.com/d/downloads.daily.png
>
> I don't think that's a coincidence and attribute a significant chunk of that to his efforts and those who wrote posts, which is why I suggested starting an official blog years ago.

The big recent spike appears to coincide with DMD being re-licensed as fully open source, as well as the GDC inclusion into GCC.

Years ago I was interested in D but considered the licensing to be a show-stopper. I've recently come back to learn it proper and try it for some small projects precisely because of the licensing change.

On Tuesday, 28 November 2017 at 23:07:32 UTC, John Gabriele wrote:
> The big recent spike appears to coincide with DMD being re-licensed as fully open source, as well as the GDC inclusion into GCC.
>
> Years ago I was interested in D but considered the licensing to be a show-stopper. I've recently come back to learn it proper and try it for some small projects precisely because of the licensing change.

I didn't download it cause someone was blogging about it ;-)

I only downloaded it because I discovered ldc2 in FreeBSD ports, and it mentioned a new langauge called D, which I had never heard of. After a bit of googling, I discovered the reference compiler was fully released as open source, under the Boost licence. (had it been GPL'd my interest would likely have stopped there, and had it only been the frontend, and not the backend, my interest would have stopped there too).

And knowing that LLVM was on board, was a really important factor for me too.