== Quote from The Anh Tran (trtheanh@gmail.com)'s article
> In my limit knowledge, D2 has invariant+pure integrated already, to
> support functional paradigm. Is there any 'feature', cooking in
> backyard, so that d2 can't be declared stable?
> This is purely speculation:
> 1. Concurrent support?
> 2. Operator overhaul?
> 3. Add more paradigm??? :D
> 4. Something to assimilate Martian people? :D

Well, this _was_ supposed to happen in April of this year b/c that's when Andrei's book was supposed to come out.  Whether that's still feasible, IDK.

On Wed, 25 Mar 2009 20:07:02 +0300, The Anh Tran <trtheanh@gmail.com> wrote:

> In my limit knowledge, D2 has invariant+pure integrated already, to support functional paradigm. Is there any 'feature', cooking in backyard, so that d2 can't be declared stable?
>
> This is purely speculation:
> 1. Concurrent support?
> 2. Operator overhaul?
> 3. Add more paradigm??? :D
> 4. Something to assimilate Martian people? :D

I believe memory model (shared/thread-local separation etc) is not fully implemented yet.
Other than that it's quite stable already and last few releases broke (almost) none of my code.

P.S. Please, start new topic by clicking "Compose", not "Reply" next time.

The Anh Tran wrote:
> In my limit knowledge, D2 has invariant+pure integrated already, to support functional paradigm. Is there any 'feature', cooking in backyard, so that d2 can't be declared stable?
> 
> This is purely speculation:
> 1. Concurrent support?
> 2. Operator overhaul?
> 3. Add more paradigm??? :D
> 4. Something to assimilate Martian people? :D

Concurrency support.

Walter Bright:
> Concurrency support.

Is it possible for you to find a sponsor willing to pay you to implement exceptions on Windows inside LLVM? I think LLVM interests enough people, and exceptions can then be used in C++ code too.
I have no idea of the amount of work required, but maybe one or two months of work may suffice to do that.

Once LLVM has them, LDC can use them and can be used on Windows too. So LDC may become the reference D compiler (and it can grow support for D2 language in some months).
LLVM isn't as efficient as GCC, but it's modern and actively developed.

(Eventually a project like dil too (D front-end for LLVM written in D) may become viable, but that's for the future).

Bye,
bearophile

On Wed, Mar 25, 2009 at 4:10 PM, bearophile <bearophileHUGS@lycos.com> wrote:
> Walter Bright:
>> Concurrency support.
>
> Is it possible for you to find a sponsor willing to pay you to implement exceptions on Windows inside LLVM? I think LLVM interests enough people, and exceptions can then be used in C++ code too.
> I have no idea of the amount of work required, but maybe one or two months of work may suffice to do that.

What makes you have any expectation that Walter has anything to do with LLVM?

Jarrett Billingsley:
> What makes you have any expectation that Walter has anything to do with LLVM?

Isn't LLVM open source? So "anyone" can help, especially if such person finds someone willing to pay for such work.
Seeing how GCC (and probably in the future LLVM too) are at the base of so much open source software in the world, I think someone may be willing to pay a developer for some time to implement some useful feature.
(Of course, everything I have said here may be stupid & wrong).

Bye,
bearophile

On Wed, Mar 25, 2009 at 5:50 PM, bearophile <bearophileHUGS@lycos.com> wrote:
> Jarrett Billingsley:
>> What makes you have any expectation that Walter has anything to do with LLVM?
>
> Isn't LLVM open source? So "anyone" can help, especially if such person finds someone willing to pay for such work.

Suuure, but do you honestly think Walter's going to drop everything he's doing and work on a project that he is currently not involved in in any way, to support a compiler that he didn't make?

It just seems very unlikely, is all.

bearophile wrote:
> Is it possible for you to find a sponsor willing to pay you to implement exceptions on Windows inside LLVM? I think LLVM interests enough people, and exceptions can then be used in C++ code too.
> I have no idea of the amount of work required, but maybe one or two months of work may suffice to do that.

Yes, Google—if it’s a student doing the implementation; LLVM is a mentoring organization in Google’s Summer of Code 2009.

—Joel Salomon