When an object gets a truncated name (and url) because the source attribute is longer than ContentSettings/ContentObjectNameLimit, php uses substr() to create the shortened version of the name.

But that is not utf8-safe, and will create invalid utf8 strings. The db will then halt with a transaction failed error and prevent such object to be stored.

As I do not think eZ wants to enable mb_* functions by default, we should fix eZNamePatternResolver::resolveNamePattern instead

(for options: see http://stackoverflow.com/questions/1967540/truncate-a-utf-8-string-to-fit-a-given-byte-count-in-php)