Re: [DotGNU]Internet C++

[Barry Fitzgerald]

Keith Poole wrote:
> 
> I was just looking at the  .GNU website
> (http://www,gnu.org/projects/dotgnu) and saw this bullet point:
> 
>     * Adapt GCC to so that it can, from all its supported languages,
>       generate portable executables for the DotGNU platform. We have
>       found that it is extremely difficult to adapt gcc to generating
>       code for stack-based VMs. So maybe the way to go is invent a new
>       (more register-based) virtual machine, and add support for it to
>       the DotGNU platform.
> 
> One thing I was wondering, couldn't we take an architecture, remove any
> restrictions that are placed on it because it's implemented in hardware,
> and use that?  The changes to gcc would be smaller.
> 
> For example, choosing the SPARC processor which is open, and partially
> stack based anyway (i.e. rolling stack frame of registers), give each
> thread of execution in the VM its own register stack, and remove the
> architected maximum of 32 frames (make it infinite) and use this?
> 
> Also, there are other open source processors which mayh be useable - but
> I don't know enough about them to comment.
> 
> Keith Poole


Forgive me if I'm misunderstanding where you're going with this, but
wouldn't that create incompatible IL between architectures?  Or, are you
referring to modeling the stack types, etc.. after those in the
processor types which means that some of the work is done for us in the
GCC backend?  If so, isn't that implemented after the fact anyway?

Please, comment further on this - I'm not sure what this would save us.

        -Barry