Re: [Gnu-arch-users] Re: new language, arch, furth, etc.

[Tom Lord]

    > From: Daniel Skarda <address@hidden>


    > >         <assoc> := "{" <assoc-binding>* "}"

    > >         <assoc-binding> := <name> "=>" <nested-value>

    >   as I understand you want to make configuration language subset
    >   of Scheme

Well, I want the datatypes to be isomorphic to something that can be
built from Scheme types in a useful (i.e. pragmatically interoperable)
way.

I'm inclined towards a syntax that a Scheme reader can read.

I'm committed to an operational model that can be implemented nicely
in Scheme but -- really -- that's not saying much.

    > (good choice, imHo :) [1]. I think that the choice of curly braces is not 
a good
    > one, because:

    >   1) R5RS declares that:
    > 
    >   [ ] { } |
    >        Left and right square brackets and curly braces and vertical bar
    >        are reserved for possible future extensions to the language

Yeah, i'm being a little arrogant there.  "Hey, I have a use for {}."

The Scheme definition is, although there are promises to fix it, a
little bit politically horked these days.   So, I'll just pretend I'm
an R^*S editor/author and let the chips fall where they may.

    >   2) What is so bad about association lists? 

You can't explain them without cons pairs.  We don't need to introduce
cons pairs.   An implementation could _use_ association lists, sure --
but I regard assocs as an extensional type that describes a physical
representation while assocs are an intensional type that describes an
abstract structure.

    >      If configuration langauge is pure subset of Scheme, you can `read' it
    >      and process it in few lines of code using any Scheme implementation
    >      (Guile, Pika, Systas, ...). You can even `read' it using any common
    >      Common Lisp implementation :)

Right.  I've thrown in some monkey wrenches like the "#undefined"
value -- but basically right.

    >      I do not think '{' '}' syntactic sugar is worth of troubles,
    >      incompatibility and writing special parsers (when there are already
    >      about 11 Scheme and 4 Lisp implementations ready for installation in
    >      Debian :)

    > Please, KISS the configuration language :)
    > 0.

yes, yes, yes.   These are (important) details that are mostly below
the current level of attention of the presentation.   I'm just
"glossing over" not "blowing off" some of these issues because I think
that no matter how they come out, it's not a big deal either way.

-t