Re: [Chicken-users] need idea for read-syntax

[felix]

On Tue, 18 Nov 2003 10:53:22 +0000, Category 5 <address@hidden> 
wrote:

> felix wrote:
>
> > Ok, shall it be #{ ... }#, then?
>
> Hmmn.  This isn't too bad for the statement-sequence case, where the {}
> even suggest the block structure of a C function, but I don't think it
> fits the typed-value case well at all (partly for the same reason).
>
> I can see Peter's point, but I think the use cases of #[ ]# are
> different enough from literal vector notation to avoid confusion.  (Then
> again, I don't have any code that writes vectors as #[1 2 3].  =)

Personally, I don't care much wether it's #{ ... }# or #[ ... ]#.
The problem with the former is that it would some changes to the compiler
(which uses it extensibly for another hackish purpose), and that would
cause a bootstrapping problem (not serious, though). The problem with the
latter is that the association with embedded C code seems a bit far- 
fetched.
Another option would of course be to drop it entirely and use `foreign- 
code'.

> Actually I'd rather see the conflation of []/() disappear and have plain
> [] used to represent vectors in Scheme, making them look less like
> second-class citizens.  But that's a whole 'nother discussion.

Absolutely. I use [] quite a lot as a replacement for normal parens,
especially when nesting gets deep:

(let-values ([(x y z) ...])
 ...)

appears to be more readable (for me) than

(let-values (((x y z) ...))
 ...)


cheers,
felix