[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: [idn] Re: stringprep and unassigned code points



At 5:35 PM -0800 10/30/01, Yves Arrouye wrote:
>You are missing my point:

Sorry, I can now see that. I thought you were talking about software, 
not people's perceptions.

>  at the time the application gets the code point
>from its environment (the operating system), the code point *is* assigned.
>It is just not considered so because of the Stringprep profile compiled in
>the application. So while this code point is perfectly fine for 1/ the user,
>2/ the people who entered a properly processed string in Server S, the net
>effect, *from the user's perspective* may be that they see some literature
>that advertise going to \U00020000.com (where \U<eight-hex> represents a
>Unicode code point), type exactly that, but cannot get there.

Ah, I see what you are saying now. Sorry for my denseness. You are, 
of course, correct. Anyone who advertises a host name that only 
exists in some newer version of nameprep, and the string that they 
advertise will come out of the newer version different than it went 
in, will be doing a disservice to any user who has an old version.

I'll try to think up some wording relating to this and send it to the 
list for inclusion in the next draft.

>My only conclusion from that was that it is not correct to say that
>applications or their supporting libraries (for Stringprep/Nameprep) will
>not need to be upgraded.

No one ever said that they did not need to be upgraded. They do not 
need to be upgraded in order to enter characters that are valid in 
stored host names. There is a big difference.

>  It may be very easy to do as I said (e.g. on an OS
>where every app uses a shared lib for that, upgrade the lib when you add
>support for the code points in the OS through a language pack or whatever).
>It may also be a nightmare for users of many applications that each use
>their own implementation.

It is only a "nightmare" for the people who enter characters that are 
mapped. Again, it would be silly for someone to advertise a name 
knowing that people with older clients could not enter it.

--Paul Hoffman, Director
--Internet Mail Consortium