On 02/18/2015 01:23 PM, Shiro Kawai wrote:
> Looks good, but I still want the document to emphasize that
> the portable program should assume string-append! can be
> no more efficient as string-append.   (Of course it's fine
> to use it on implementations on which you know string-append!
> is more efficient.  It's just that you can't count on it
> in general.)

Right - but is that any different from string-set! - i.e. you can't
count on string-set! being significantly more efficient that
string-append/string-copy. And on implementations where string-append!
requires a copy, there string-set! would have the same problem.

So the place, if any, to mention this warning would seem to be R7RS ....

On 02/18/2015 02:02 PM, xxxxxx@ccil.org wrote:
> BTW, the non-first arguments of string-append! presumably have to
> be strings, not arbitrary Scheme values.

The draft says:

     A value can be a character or a string.
--
	--Per Bothner
xxxxxx@bothner.com   http://per.bothner.com/