(see also to the bug MAGNOLIA-3008)

Magnolia allow a very reduced set of characters to be used for page names, replacing everything non-ascii char by "-". This is pretty "aggressive", since there is no reason to not support for example accented chars or similar, which are perfectly valid for jcr node names.

Especially for the always increasing importance of URLs for SEO this could be a relevant problem.
For example if you try to create a page with the name "pietà" the resulting url is "piet-" which is definitively bad. Also changing the name to "pieta" is better but not satisfying in terms of SEO.
For asian languages this is even worst, you can't use speaking urls at all, since all the characters will not be supported. Trying to create a page named "モクレン" (magnolia in japanese) results in a page named "----.html" which is definively not user/seo friendly...

A preliminary check showed that supported non-standard characters should not be so hard (before the MAGNOLIA-3008 related changes actually existing content with extended chars was already readable by Magnolia).

We should probably review all the character-killing methods (like Path.getValidatedLabel() and similar) in order to admin a richer set of chars.
If somebody think this could be a big change for a minor version we may simply add a new system property like "magnolia.allowutf8urls" that can be enabled explicitly (and that we can enable by default in Magnolia 5)