janderson Wrote:

D gets popularity but not enough.
IMHO the main factor is the connection with other languages.
The only connection D has is via C interface which is not enough.

I think D should support COM and C API in a simple way. Even by native language support.

This is need because D standard library is rare, but with simple COM support, for example, D can use many Windows objects.

What about porting Boost.Python for D as starting point ?
Thus Python can easily be connected to D program and this can increase language usage.

And of course DMD should be replaced by GDC , because 2 compilers do porting problem.
GDC has a big advantage because the back end is same for GCC so it can get all optimizations automatically.

NN wrote:

> janderson Wrote:
> 
> D gets popularity but not enough.
> IMHO the main factor is the connection with other languages.
> The only connection D has is via C interface which is not enough.
> 
> I think D should support COM and C API in a simple way. Even by native language support.
> 
> This is need because D standard library is rare, but with simple COM support, for example, D can use many Windows objects.
> 

D supports COM and interfaces with C just fine.

> What about porting Boost.Python for D as starting point ?
> Thus Python can easily be connected to D program and this can increase
> language usage.

There is already PyD which is probably going to kick Boost.Python's ass when it matures.

> 
> And of course DMD should be replaced by GDC , because 2 compilers do porting problem. GDC has a big advantage because the back end is same for GCC so it can get all optimizations automatically.

NN wrote:
> janderson Wrote:
> 
> D gets popularity but not enough.
> IMHO the main factor is the connection with other languages.
> The only connection D has is via C interface which is not enough.

AFAIK pretty much every programming language "speaks" C. It's basically the lingua franca of the programming world.

> I think D should support COM and C API in a simple way.
> Even by native language support.
> 
> This is need because D standard library is rare, but with simple COM support, for example, D can use many Windows objects.

Are you aware of the last section of http://www.digitalmars.com/d/interface.html ?

> What about porting Boost.Python for D as starting point ?
> Thus Python can easily be connected to D program and this can increase language usage.

http://dsource.org/projects/pyd ?

> And of course DMD should be replaced by GDC , because 2 compilers do porting problem.
> GDC has a big advantage because the back end is same for GCC so it can get all optimizations automatically.

Having multiple implementations is a good thing. Some people (and companies) have a preference for open-source or commercial software. Some might also be hesitant to use a language with only a single compiler available.
A show-stopping bug in one compiler might also be worked around by switching compilers.

== Quote from Frits van Bommel (fvbommel@REMwOVExCAPSs.nl)'s article
> NN wrote:

(snip)

> > I think D should support COM and C API in a simple way. Even by native language support.
> >
> > This is need because D standard library is rare, but with simple COM support, for example, D can use many Windows objects.
> Are you aware of the last section of http://www.digitalmars.com/d/interface.html ?

I think NN is complaining about how difficult it is to use existing COM classes from D. It is a lot of trouble right now.

I think that Juno (http://www.dsource.org/projects/juno) has the potential to make
COM a lot simpler in D, but it hasn't officially caught up to D 1.0 yet, and I
like to work with more current versions of the DMD compiler. So I'm hoping that
Juno's author returns and updates Juno. In the meantime, another user (oldrev)
came up with an update
(http://www.dsource.org/projects/juno/wiki/UserContributions). I plan to use
oldrev's work to come up with some more examples, but my plans often get sidetracked.

jcc7

Frits van Bommel Wrote:

> NN wrote:
> > janderson Wrote:
> > 
> > D gets popularity but not enough.
> > IMHO the main factor is the connection with other languages.
> > The only connection D has is via C interface which is not enough.
> 
> AFAIK pretty much every programming language "speaks" C. It's basically the lingua franca of the programming world.
It is but still there is much C++.

> 
> > I think D should support COM and C API in a simple way. Even by native language support.
> > 
> > This is need because D standard library is rare, but with simple COM support, for example, D can use many Windows objects.
> 
> Are you aware of the last section of http://www.digitalmars.com/d/interface.html ?
Yes, but it doesn't helps too much.
How the proxy-stub generated ?
VC supports #import which simplifies your life a lot, can D do the same ?

What happens if the interface in the top isn't derived from IUnknown:
interface A { . .}
interface B : A, IUnknown { .. }


> 
> > What about porting Boost.Python for D as starting point ?
> > Thus Python can easily be connected to D program and this can increase language usage.
> 
> http://dsource.org/projects/pyd ?
+1
Nice, didn't know about it :)
Anyway it is a good starting point but not the end.


> 
> > And of course DMD should be replaced by GDC , because 2 compilers do porting problem.
> > GDC has a big advantage because the back end is same for GCC so it can get all optimizations automatically.
> 
> Having multiple implementations is a good thing. Some people (and
> companies) have a preference for open-source or commercial software.
> Some might also be hesitant to use a language with only a single
> compiler available.
> A show-stopping bug in one compiler might also be worked around by
> switching compilers.