Bootstrapping language implementations without C

[David Hopwood]

Samuel Bronson wrote:
> On 15/05/05, David Hopwood <david.nospam.hopwood at blueyonder.co.uk> wrote:
>>Samuel Bronson wrote:
>>>On 5/9/05, Pupeno <pupeno at pupeno.com> wrote:
>>>
>>>>We may see the time when writing a static-language is no longer worth it,
>>>>because the dynamic-language are so much better... and we might eventually
>>>>see C die, leaving the field to newer languages... [snip!]
>>>
>>>However shall we compile our interpreters then?
>>
>>Using MLRISC for example (http://www.cs.nyu.edu/leunga/www/MLRISC/Doc/html/).
>>It is written in Standard ML, and compiles various high-level languages to
>>various machine ISAs. It does not depend on C at all, only on the fact that
>>some Standard ML implementation exists for some machine.
> 
> But then I would need to write in SML. This might not be too bad with
> monads, which I hear are coming to some ML or other,

MetaML? <http://www.cse.ogi.edu/PacSoft/publications/phaseiiiq9papers/dsl.pdf>

> but it might have been the other,

MLRISC is developed using SML/NJ, and MetaML is also based on SML/NJ.
Running MLRISC on MetaML would probably just work.

> but even so the syntax seems rather odd.  I also don't know what I would do
> in a  HM-based typesystem without type classes. In any case, SML is
> statically-typed.

<shrug>. Surely these factors don't outweigh the value of using a well-tested
framework that already has a Smalltalk front-end, and back-ends for all the
common processor architectures?

-- 
David Hopwood <david.nospam.hopwood at blueyonder.co.uk>