--- Begin Message ---
On Wed, 2009-02-18 at 20:09 -0500, John Cowan wrote:
> I'm not an authority on that. R5RS requires only the ASCII
> character repertoire; various Schemes have supported Unicode
> in various ways and to various degrees.
>
> It's been a vexed question whether R5RS and earlier RnRS documents
> imply that strings must provide O(1) random access to the characters
> that constitute them. It is not so in R6RS.
I've made arguments about why I think "no worse than O(log n)" is
probably a better choice than "O(1)." It's because Log-n allows strings
to be implemented as ropes and permits much faster length-changing
operations, data sharing of common substrings, O(1) copying, etc.
I would be happy for any new standard to continue not to promise O(1)
character access.
But this is a language issue (or a library issue), and not a
procedure issue. ie, you're talking about something that the
editors should take up, not something that the members of the
steering committe should take up.
Bear
--- End Message ---
_______________________________________________
r6rs-discuss mailing list
[email protected]
http://lists.r6rs.org/cgi-bin/mailman/listinfo/r6rs-discuss
