On Feb 21, 2013, at 09:38 , Hans-Peter Locher <hans-peter.locher@inquant.de> wrote:

In a distributed application, you cannot always ensure that you have each component "under control". The url is the most basic interface to a web application.

In Plone the id (and therefore the url) is generated from user's title input. For a title unicode is perfectly fine and needed (especially in unicode countries like germany).

If Plone/Zope would generate utf-8 ids by default you'll get utf-8 urls from the users generating the content immediately.

This will lead to unpredictably errors (as most urls will still work as they don't contain utf-8 but some do) in the distributed applications.

What kind of errors are you thinking of?

Another issue might be that some developer tools might stop working / or only work in latest versions

E.g.

Python's urllib and xmlrpclib Selenium IDE windmill jmeter …

All of these will still work fine. One thing to realise is that an UTF-8 URL itself is still an ASCII thing: it uses %-quoting for non-ASCII (and some other) characters. For example my local test site has a URL which looks like http://localhost:8080/Pløne/ in my browser but it really is http://localhost:8080/Pl%C3%B8ne/ . To get there Zope uses urllib.quote() internally, for example in the implementation of the absolute_url() method.

In my opinion having utf-8 urls should be possible in Zope/Plone. But not by default.

I have no problem with making that configurable. Wichert.