News article, Microsoft releases Clang with Microsoft CodeGen:

http://blogs.msdn.com/b/vcblog/archive/2015/12/04/introducing-clang-with-microsoft-codegen-in-vs-2015-update-1.aspx

The interesting bit is at the end:

"
Clang with Microsoft CodeGen isn't just a private fork of the open-source Clang compiler. We'll be contributing the vast majority of the Clang and LLVM changes we've made back to the official Clang and LLVM sources. The biggest of these changes is support for emitting debug information compatible with the Visual Studio debugger
"

With these developments, one asks again, is it wise to spend any more time working and using the Digital Mars backend for D?...

-- 
Bruno Medeiros
https://twitter.com/brunodomedeiros

On Monday, 7 December 2015 at 11:26:27 UTC, Bruno Medeiros wrote:
> News article, Microsoft releases Clang with Microsoft CodeGen:
>
> http://blogs.msdn.com/b/vcblog/archive/2015/12/04/introducing-clang-with-microsoft-codegen-in-vs-2015-update-1.aspx
>
> The interesting bit is at the end:
>
> "
> Clang with Microsoft CodeGen isn't just a private fork of the open-source Clang compiler. We'll be contributing the vast majority of the Clang and LLVM changes we've made back to the official Clang and LLVM sources. The biggest of these changes is support for emitting debug information compatible with the Visual Studio debugger
> "
>
> With these developments, one asks again, is it wise to spend any more time working and using the Digital Mars backend for D?...

Well not everyone uses visual studio. It may be of help to other win guys though. Not me ATM. So yes it is soo oo worth it IMO.

On Monday, 7 December 2015 at 14:22:15 UTC, Karabuta wrote:
> On Monday, 7 December 2015 at 11:26:27 UTC, Bruno Medeiros


_________________
NOOR

On Monday, 7 December 2015 at 11:26:27 UTC, Bruno Medeiros wroçte:
> ...
> With these developments, one asks again, is it wise to spend any more time working and using the Digital Mars backend for D?...

The problem are the companies to work, where I live today the main jobs are for C# or
Java, I'll not bother to mention of course HTML, JS, PHP...

And what's the relation with you asked? Well many companies here have partnership with Microsoft and they use their products like VS, so I don't think it's wise.

Matheus.

On Wednesday, 9 December 2015 at 14:35:29 UTC, Bubbasaur wrote:
> The problem are the companies to work, where I live today the main jobs are for C# or
> Java, I'll not bother to mention of course HTML, JS, PHP...
>
> And what's the relation with you asked? Well many companies here have partnership with Microsoft and they use their products like VS, so I don't think it's wise.

Hm, I kind of agree with Bruno, and I don't really understand why dropping a homegrown backend would not be wise? It probably keeps the language from evolving.

If clang and gcc become the de-facto standard compilers then C++ will gain a new edge because then the shared subset of extensions that clang/gcc share will become portable and adoption of new standards will be sped up.

On Wednesday, 9 December 2015 at 15:29:45 UTC, Ola Fosheim Grøstad wrote:
> Hm, I kind of agree with Bruno, and I don't really understand why dropping a homegrown backend would not be wise?

Oh nevermind, I misread, thought you meant it wouldn't be wise to drop it. I get it now ;-).

On Monday, 7 December 2015 at 11:26:27 UTC, Bruno Medeiros wrote:
> With these developments, one asks again, is it wise to spend any more time working and using the Digital Mars backend for D?...

I asked this very question about a year ago. The thread is here:

http://forum.dlang.org/post/mjwitvqmaqlwvoudjoae@forum.dlang.org

On Monday, 7 December 2015 at 11:26:27 UTC, Bruno Medeiros wrote:
> News article, Microsoft releases Clang with Microsoft CodeGen:
>
> http://blogs.msdn.com/b/vcblog/archive/2015/12/04/introducing-clang-with-microsoft-codegen-in-vs-2015-update-1.aspx
>
> The interesting bit is at the end:
>
> "
> Clang with Microsoft CodeGen isn't just a private fork of the open-source Clang compiler. We'll be contributing the vast majority of the Clang and LLVM changes we've made back to the official Clang and LLVM sources. The biggest of these changes is support for emitting debug information compatible with the Visual Studio debugger
> "
>
> With these developments, one asks again, is it wise to spend any more time working and using the Digital Mars backend for D?...

Walter has decades invested in his backend, he won't even look at code for other compilers.  He's still working on his dmd backend, just added DWARF exception-handling support.  Dmd is still the fastest to compile and provides reasonably good code generation, though not the best, so dmd still has use as a fast development compiler.

Let's see, did I miss a reason?  These are all the ones I've read on the forum in the past.

On Thursday, 10 December 2015 at 01:09:30 UTC, Joakim wrote:
> Let's see, did I miss a reason?  These are all the ones I've read on the forum in the past.

But the real question is whether it is a strategic good move?

Go is the only language now that use its own backend and they loose performance over it, and get bad comments for it, but they get to tailor it to a reasonable GC so it has some strategic value.

Rust recently announced that Mozilla is going to include Rust code in their products in 2016. So they are committed.

The science people seem to rally behind Julia JIT, and a JIT and mindshare is important in that field.

With Swift on Linux the ARC approach becomes less attractive for other languages as you put yourself up for direct comparison. If Swift can get reasonable performance on Linux and Android then they will take a fair marketshare real fast because of tooling and portability, both on mobile and even on web servers.

In this crowded "close to production ready" landscape competition becomes more fierce. I think languages like Swift going cross platform will create trouble for languages like Nim and D.

On Thursday, 10 December 2015 at 02:22:34 UTC, Ola Fosheim Grøstad wrote:
> On Thursday, 10 December 2015 at 01:09:30 UTC, Joakim wrote:
>> Let's see, did I miss a reason?  These are all the ones I've read on the forum in the past.
>
> But the real question is whether it is a strategic good move?
>
> Go is the only language now that use its own backend and they loose performance over it, and get bad comments for it, but they get to tailor it to a reasonable GC so it has some strategic value.
>
> Rust recently announced that Mozilla is going to include Rust code in their products in 2016. So they are committed.
>
> The science people seem to rally behind Julia JIT, and a JIT and mindshare is important in that field.
>
> With Swift on Linux the ARC approach becomes less attractive for other languages as you put yourself up for direct comparison. If Swift can get reasonable performance on Linux and Android then they will take a fair marketshare real fast because of tooling and portability, both on mobile and even on web servers.
>
> In this crowded "close to production ready" landscape competition becomes more fierce. I think languages like Swift going cross platform will create trouble for languages like Nim and D.

It would be nice to have a D JIT that is fast as others and can be easily used in a D app and interface with it's host without too much work.

The more D can do to cover as much ground as it can well, the more attractive it is, right?