Paolo Carlini wrote:
Isn't it possible to use the same strategy? That is break the
string into chunks separated by \0, then transform each chunk
with strxfrm (again, untested code):
Interesting.
Do you believe that proceeding in this way the corresponding
results are always consistent?
I mean, collate::transform and then plain string compare always
give the same result of collate::compare?
Yes, I think I can give a formal proof of this if anybody is
interested (hint: split each string x into zero-terminated
substrings x1, x2, ... xn, then consider the results for each
substring xi), but I would like to at least compile this before
giving a definite answer :-)