[lsb@lsb.org: [EAI] (summary) display of RightToLeft chars in localparts and hostnames]

Soobok Lee lsb at lsb.org
Thu Dec 7 10:31:33 CET 2006


On Thu, Dec 07, 2006 at 10:12:31AM +0100, Harald Alvestrand wrote:
> Soobok Lee wrote:
> >
> >My new suggestion is that: stringprep processes
> >  <RLE>D N A R T S E V L A<PDF> ==> ALVESTRAND 
> >  <LRE>YOD HE WOW HE<PDF> ==> HE WOW HE YOD ( in Hebrew)
> >  instead of just deleting or prohibiting <RLE> and <LRE>.
> >
> >How do you think about this "Just delete with reordering"?
> >It won't complicate stringprep algorithms so much.
> I suppose it's possible to execute the whole bidi algorithm of UAX#9 and 
> re-code the result as some kind of "normalized RTL". Is there a 
> normalization algorithm for bidi in Unicode?
> 
> But I don't see that it's reasonable to expect EVERY IDNA implementation 
> to do this - complexity is WAY higher than for many other things.
> 
> If we make a clear separation between "allowed characters on the wire" 
> and "advice to implementors on how they can help people recover from 
> weird-encoding errors", this may go into the latter part.

Yes. I agree with your classification. 
the reason why i suggested this is  to enable:

 stringprep( FirstRTL(RLE) )==> FirstRTL     without error,
  in the display context of  FirstRTL(RLE).(PDF)SecondRTL(RLE).(PDF)com example.

 RL* LR* won't be allowed on the wire even after stringprep, as you said.
 IDN Registrants won't enter (RL*), i guess.

Soobok



More information about the Idna-update mailing list