Slate: extreme optimization

[Brian T. Rice]

Lyndon Tremblay wrote:

> So, this is called Slate, and these .slate files, is it just me that sees
> this inconsistance? Rather, too much consistance if the future expects less
> of it? The reason I didn't realise that Slate "isn't the only language" for
> ... Slate.

Keep in mind that Pidgin is something of a compromise to speed up 
bootstrapping, by being C-translatable. Eventually we want to have 
language-modes which are less at-conflict with Slate semantics and fit 
into the language itself (and we want to skip C translation even so).

> Clarification on current protocols regarding multilingual expansion? Pidgin
> files are also .slate, using the same emacs syntaxer (leveraging off
> smalltalk-mode.el), etc etc.

I don't understand. What do you mean by "current protocols regarding 
multilingual expansion"? Do you mean human or programming languages? The 
former is addressed in the recent thread on Unicode (well, partly). The 
latter is addressed by the fact that Slate is a platform that can be 
compiled-into without the same syntax. A lisp style of syntax has 
crossed our minds sometimes, for example; most of the common syntaxes 
might be interesting to others, but not to us... we are mostly 
interested in abstracting the syntax we have.

> This is quite interesting though, but how to avoid the .NET-isms of multiple
> and transparent language support? If there are any.

We can avoid the ambiguity by annotating source trees or packages with 
the "mode" or "dialect" of the contents, and dispatching on that. (Part 
of the dispatching would occur in tools, of course, to help indicate 
subtly but clearly what mode you're in.)

We actually have an inverted approach compared to .NET, where they have 
several syntaxes for the same semantics, and we are offering different 
modes and manners of usage for the same language.

> --Lyndon
> 
> 
>>Nothing stops you from embedding a more efficient numerics language in
>>Slate. We already do exactly this with Pidgin for writing the VM! We are
>>a system, NOT just a language. A system can encompass multiple
>>languages, and to some extent Slate is a system of multiple languages
>>and may be moreso in the future.
>>
>>Lee
>>
>>On Thu, Jul 29, 2004 at 05:42:12PM +0200, Waldemar Kornewald wrote:
>>
>>>You only talk about the tasks that Slate fits best. Slate is easy to
> 
> learn and use, powerful, etc., but when you want to write a fast embedded OS
> for your multimedia-station in your living-room or when you want to write a
> video codec, you will not even get a fast FFT algorithm.
> 
>>>You don't know how hard it would be to implement that, do you? :)
>>>
>>>Bye,
>>>Waldemar
>>>

-------------- next part --------------
A non-text attachment was scrubbed...
Name: water.vcf
Type: text/x-vcard
Size: 208 bytes
Desc: not available
Url : /archives/slate/attachments/20040729/2b4d6c56/water.vcf