[mono-vb] Future of Mono's VB.NET

Rodrigo B. de Oliveira rodrigobamboo at gmail.com
Thu Apr 20 11:38:15 EDT 2006


On 4/20/06, Miguel de Icaza <miguel at novell.com> wrote:
> Hello,
>
> > Let me put another possibility on the table. Why not base the new
> > VB.NET compiler on the Boo infrastructure?
> >
> > This would not require a fork at all but improvements and refactorings
> > to a shared code base on a ongoing basis. There are already two
> > languages based on the infrastructure (boo itself and the scripting
> > language for the Unity Game Engine - UnityScript).
> >
> > I'd love to see that collaborative development happening.
>
> But isn't Boo still missing generics support?
>

Not for long :)

> It just feels to me that VB.NET and C# are close enough that a compiler
> forked from gmcs is probably easier to do than reusing Boo, specially
> considering the generic components.
>

Well, Boo was already described as a mix of VB.NET and C# by some.

We need a different parser in any case. Just by integrating this new
parser into the boo pipeline would get us a long way. In fact Daniel
Grunwald integrated the #develop VB.NET parser into boo some time ago
so it's possible to write some kind of VB code already with boo. The
semantic differences were not implemented but that would be the same
work for boo or gmcs with one nice difference: we would be sharing
code!

> Rafael did mention this Boo proposal, but I do not know enough about the
> internals of Boo or what Boo exposes that would make this easier.
>

Boo architecture is open enough so you can plug different parsers,
different type system rules, overloading resolution schemes and such.
Don't get me wrong it would still take a lot of work to get that
openess to a point where we will say: beatiful! But collaboration can
certainly gets us there and it looks like a fun ride from where I
stand.

> Finally, the other downside is that integrating the patches from mbas
> into a new fork is mostly a straightforward task, while basing it on Boo
> sounds like writing code from scratch.
>

Forking is generally a dead-end for code sharing. That's what happened
to mbas. That's what will eventually happen to mbas2 I guess. At some
point the two versions just diverge so much that the gap just gets too
scary.

What will it be, the blue pill or the red pill? :)
Rodrigo


More information about the Mono-vb mailing list