On 22 Oct 2013, at 15:49, "[email protected]" <[email protected]> wrote:

> I'd say you can encode @ with % in the URL (but it is not needed in any case) 
> indeed but in the rendered output? (Do you mean in hrefs="xxx" ?)
>  
> Some things must be encoded indeed:
>  
> http://en.wikipedia.org/wiki/Character_entity_reference
> http://en.wikipedia.org/wiki/List_of_XML_and_HTML_character_entity_references#Character_entity_references_in_HTML
> http://www.w3schools.com/tags/ref_urlencode.asp
>  
> As '@' is in the ASCII set, why should it be encoded?

Because that is what the spec says:

http://en.wikipedia.org/wiki/Percent-encoding#Types_of_URI_characters

These characters, 

!       *       '       (       )       ;       :       @       &       =       
+       $       ,       /       ?       #       [       ]

have to be percent encoded in external (printed) representations of URLs, ASCII 
or not.

This has nothing to do with HTML or XML.

ZnPercentEncoder uses the following

A       B       C       D       E       F       G       H       I       J       
K       L       M       N       O       P       Q       R       S       T       
U       V       W       X       Y       Z
a       b       c       d       e       f       g       h       i       j       
k       l       m       n       o       p       q       r       s       t       
u       v       w       x       y       z
0       1       2       3       4       5       6       7       8       9       
-       _       .       ~

as its safe set (characters that do not have to be encoded), just like the spec 
says.

I could be wrong, but I don't think so.

Sven

> URL Encoding of @ is %40 indeed but digits also have encodings. Not a reason 
> to use them :-)
>  
>  
>  I'd say that the issue is in ZnClient, basically defaulting to a capability 
> that is linked to server interpretation of URLs.
> There is no reason why the URL you use must be associated with 
> authentication, only that mainstream servers can do that with mod_auth or 
> something.
>  
> Other than that it is just a pathinfo string.
>  
> A ZnClient option saying 'enableAuthority:' maybe? I do not think that a 
> better parsing of options would solve the situation as someone may want to 
> have the same scheme but for other purposes.
>  
> Phil


Reply via email to