[cosmo-dev] Unicode support in 0.11

[Travis Vachon]

Hi folks

Wanted to follow up with some conclusions and next actions regarding  
the character support I've been working on for the last week or so.

Commits 6333-6341 consist of a number of fixes and tests that support  
and enforce (client side) the following policy for usernames:

(a) Usernames must be between 3 and 32 characters; may contain any  
Unicode character in the following range of unicode code points: [#x20- 
#xD7FF] | [#xE000-#xFFFD] EXCEPT #x7F or #x3A

In English, this means: All non-whitespace (but including space), non- 
control characters in the Basic Multilingual Plane excluding the  
surrogate blocks, 0x7F (DELETE), and 0x3A (COLON).

0x7F and 0x3A are currently excluded because they are not allowed in  
Basic Authentication username tokens. The decision to not support  
characters outside of the Basic Multilingual Plane was discussed here:

http://lists.osafoundation.org/pipermail/cosmo-dev/2007-November/005363.html

The rest of the exclusions (control characters and whitespace other  
than " ") are either already present according to the Cosmo Management  
Protocol or seem self-explanatory, please let me know if they aren't.

I have not made any changes on the server side to enforce this policy.  
Currently, according to the CMP spec:

http://chandlerproject.org/Projects/CosmoManagementProtocol

the username policy we enforce server side is:

Usernames must be between 3 and 32 bytes; may contain any non- 
whitespace character as well as ' '.


If there are no objections to the username policy I suggest in (a),  
I'll file a server bug to get the server logic in place.

Thanks!

-Travis