Re: T.61 -> Unicode conversion

[Howard Chu <hyc@highlandsun.com>]

Hallvard B Furuseth wrote:

> Also,
>
> I'm pretty sure that when I used X.500 and T.61 (10+ years ago?), at
> least the T.61 Underline combining character could be combined with
> other combining characters.  I don't know if Unicode allows that; the
> strprep description of doing so at the end of Appendix A does not.

You're right, the notes I read state that the Underline can be combined 
with any other graphic character. I'll have to double-check how Unicode 
deals with this.

> Also, I may remember wrong but there seems to be a lot more valid
> character combinations (accent character + other character) than there
> was in the T.61 I am used to.  Does the strprep algorithm accept any
> T.61 character combination which can be mapped to Unicode, whether or
> not the combination is valid in T.61?  If so, there are strings which
> Stringprep will incorrectly will treat as valid.

Appendix A doesn't specify that invalid combinations should be detected. 
Perhaps it should, but it seems unfriendly to break any software that 
might be using these sequences.

> I don't know where to find T.61 so I can check this, though.  At least
> not on a Sunday:-)

Considering that the CCITT withdrew this spec years ago, it's not 
surprising. It also has never been clear which version T.61 to use, as 
that spec evolved quite a bit before being withdrawn. (e.g., 
http://www.mail-archive.com/openca-users@lists.sourceforge.net/msg04365.html

Appendix A only considers iso-ir-103 as listed in RFC1345 but there's 
really no way to know how many implementations out there did things 
differently.

--
  -- Howard Chu
  Chief Architect, Symas Corp.       Director, Highland Sun
  http://www.symas.com               http://highlandsun.com/hyc
  Symas: Premier OpenSource Development and Support