[SVN] r10527

[Allan Odgaard]

Use mblen_l() to find number of code points in UTF-8 string instead of going via CFString.

The length returned from CFString was number of UTF-16 characters, so it would previously return a length of 2 for a surrogate pair, where mblen() will correctly result in 1. Diacritics though will still result in a star per combining character.

Though while testing this, the password dialog itself doesn’t handle neither surrogates nor diacritics correctly (when displaying these as stars).

Changed:
U   trunk/Tools/tm_interactive_input/src/dialog.c