> The analogous behavior would be for generate-temporaries identifiers to
> be given the lexical context of where they are placed in a syntax
> template (if/when they are).  But because of what generate-temporaries
> identifiers are intended for, there's no reason for them to have any
> lexical context; they can't refer to anything except introduced bindings
> to themselves.
Perhaps generate-temporaries should be required to produce identifiers
with an empty context then?

On Fri, Jun 19, 2009 at 7:53 PM, Derick
Eddington<[email protected]> wrote:
> On Fri, 2009-06-19 at 18:39 +1000, Ramana Kumar wrote:
>> This section 
>> http://www.r6rs.org/final/html/r6rs-lib/r6rs-lib-Z-H-1.html#node_toc_node_sec_12.7
>> begins by saying you can introduce a fixed number of identifiers by
>> simply naming them, but when you don't know how many identifiers you
>> will need you can use generate-temporaries. Now when you introduce
>> temporary identifiers by simply naming them, they pick up the context
>> of the transformer. So it seems there's an argument that
>> generate-temporaries should behave analogously... (which would break
>> Aziz's original definition of unbound?).
>
> The analogous behavior would be for generate-temporaries identifiers to
> be given the lexical context of where they are placed in a syntax
> template (if/when they are).  But because of what generate-temporaries
> identifiers are intended for, there's no reason for them to have any
> lexical context; they can't refer to anything except introduced bindings
> to themselves.
>
>> What would be missing is a
>> way to get at an empty context.
>
> I'd suggest: (datum->syntax #F <datum>) be made possible.  But the only
> reason for an empty context is to do the free-identifier=? hack.  With a
> primitive identifier-bound?, none of this is necessary.
>
>> Is there any way to remove define and
>> syntax from the context of empty-ctxt in (library (foo) (export
>> empty-ctxt) (import (only (rnrs) define syntax)) (define empty-ctxt
>> (syntax here))) ?
>
> No.
>
> --
> : Derick
> ----------------------------------------------------------------
>
>

Reply via email to