Re: WHY we need do "Divide the sequence into segments"

dongxiaoli <[email protected]> wrote:

> "The "*protected code points" are 0..40, 5B..60, 7B..7F (in other     
> words, those corresponding to ASCII characters other than letters and 
> digits). *" is from draft-hofmann-imaa-03.txt.                        

Oops, thanks for catching that.  The prose describes the actual
intention.  The number ranges should be 0..2F, 3A..40, 5B..60, 7B..7F.

> Have anybody can tell me WHY we need do "*Divide the sequence 
> into segments*" in the ToASCII process  as described in the
> IMAA draft (draft-hofmann-imaa-03.txt)??

Before internationalized local parts can be safely created in a mail
domain, a couple questions need to be considered:

 1) Are local parts case-sensitive within this domain?

 2) Are local parts structured within this domain, using delimiters
    other than protected characters?

If the answer to either question is "yes", then internationalized local
parts will not work correctly in the domain unless the MTAs for the
domain are upgraded to understand the IMAA encoding.

However, if the answer to both questions is "no", then there is no need
to upgrade the MTAs.  Internationalized local parts can be created in
the domain and they will work correctly.

The concept of protected characters and segmentation is designed to
make it effortless to support internationalized local parts in the vast
majority of real mail domains.

The typical examples are domains that support user+tag@domain, or
which treat owner-listname@domain specially.  Because plus-sign and
hyphen-minus are protected characters, it doesn't matter whether
the joining/splitting of the components is performed on the ASCII
form or the Unicode form, because the result will be the same either
way.  If IMAA did not have segmentation, any software that dealt with
prefixed/suffixed local parts would have to be careful about the proper
order of conversions, and would therefore have to be IMA-aware.

AMC