Comment 4 for bug 371723

make_name is using a regular expression to match against non-letter
characters. I believe that regular expression library, being unicode aware,
is respecting the unicode definition of letter characters--Ie not just US
Ascii. This is worth testing, though.

We really are only trying to strip so called non-letter, ie symbol and
whitespace characters in make_name, while leaving letters, in whichever
alphabet they are defined.

Since these names are used to construct urls, though, it's probably worth
double checking that browsers actually do the right thing with unicode in
urls. I'm not entirely sure where this issue stands.

I would propose that *if* make_name really is only stripping non-letter
characters as defined above and *if* browsers can support unicode urls, that
we continue stripping those characters and show the user an error if that
creates a blank name. If one of these turns out not to be true, though,
then we should use the ascii encoding mentioned above.

Is that reasonable?

Chris