Bundling vs Mapping

[Patrik Fältström]

On 25 feb 2009, at 17.30, Erik van der Poel wrote:

> The question here is what exactly you consider to be a "bug" in
> current URI handling in the area of IDNA.

The main reason why I really want to separate domain names from URIs  
is that specifically URIs using the HTTP scheme do have certain  
characters that start the URI ("http://") and some that end the domain  
name part ("/"). These might, I claim by just having used some  
browsers, and without knowing any such script, create a specific  
context in which left to right scripts behave in a specific way.

If you just look at the domain name by itself, it can exist in both a  
R2L and L2R context, and by doing that the result when rendering the  
domain name as a string is "interesting".

So, the few tries I have done, once again -- without knowing arabic  
--, on my mac show that rendering domain names is hard. Knowing just a  
bit about context make a huge difference.

You can see my tests in a simple text editor on the mac, where I  
swapped over all directionality of the strings on the following URIs,  
and you might understand why I am worried. Regardless of whether my  
examples makes sense or not. Once again, I do not know Arabic.

http://stupid.domain.name/node/681
http://stupid.domain.name/node/682
http://stupid.domain.name/node/683

The first one of these made me wake up.

And, because of this, I think we have to keep apples separated from  
oranges, although we know we use both in a fruit salad.

    Patrik