[R6RS] draft Unicode SRFI

[Marc Feeley]

On 2-Jul-05, at 8:01 AM, Matthew Flatt wrote:

> At Sat, 2 Jul 2005 05:56:58 -0600, Matthew Flatt wrote:
>
>> At Fri, 1 Jul 2005 12:17:38 -0400, Marc Feeley wrote:
>>
>>> Fine.  In any case, I think we are close enough to agreement to let
>>> the SRFI process "do its thing".  So please send the Unicode SRFI to
>>> the SRFI editors and let the fun begin...
>>>
>>
>> Done (with one last round of edits for string escapes).
>>
>
> Undone. :(  I forgot to deal with the reference-implementation  
> section.

Don't put too much time into this.  A lot of what's specified has to  
be primitive (how can you define char->integer portably?  I don't  
think we should implement "read"... etc).

On the \<eol> and \<space> string escapes, I think our design is  
problematic because it will be hard to distinguish \<space><eol> and  
\<eol> when reading code, and these cases lead to very different  
results, i.e. compare these two string literals (the first has a  
\<eol> on the first line and the second has \<space><eol>):

    "hello\
    \ world"  => "hello world"

    "hello\
    \ world"  => "hello \n    world"

One way to fix this is to forbid <eol> in strings, except as part of  
a \<eol> escape.  Kent: does your objection to this restriction still  
stand given the inclusion of here strings that allow <eol> in the  
content?

If we can't find a solution to this, I think we should eliminate the  
\<eol> and the \<space> escapes from the SRFI.

Marc