On 11/19/14, 10:30 PM, Andrew Sullivan wrote:
On Wed, Nov 19, 2014 at 08:45:28PM -0700, Peter Saint-Andre - &yet wrote:
Nemo is proposing a profile for nicknames (with IoT or device use cases in
mind). We already have a profile for nicknames (with chatroom or chat use
cases in mind). I am suggesting that we combine the two because profiles
must not be multiplied beyond necessity.
I agree. I certainly understand the use case, but we've already seen
that we have some trouble due to the desire for multiple profiles.
I wonder though if we mightn't take a page from IDNA's playbook here.
One of the things IDNA did was to say, "Do some locale-sensitive
mapping before you get to the algorithm." Couldn't this be a case for
that? That is, half-width mapped to full width or something _before_
even starting on the string?
Well, that's what the width-mapping rule is supposed to take care of,
and it's applied first in PRECIS.
But I am thinking of other confusing strings, beyond strings with
half-width characters.
For example, in a chatroom context, we don't want someone to spoof a
user HenryIV (with last two characters U+0049 LATIN CAPITAL LETTER I and
U+0056 LATIN CAPITAL LETTER V) by creating a user HenryⅣ (with last
character U+2163 ROMAN NUMERAL FOUR).
Similarly, in an IoT context, does a user really need to have separate
devices named TabletIV and TabletⅣ or whatever? Doesn't that open the
potential for user confusion and security vulnerabilities? ("Strange,
TabletIV shows up twice, I'll just choose the first one and log in
there... oh, this is even stranger, now I can't log into any of my
tablets...")
Using the FreeformClass enables expressiveness but at the price of
potential confusion. The nickname profile attempts to reduce that
confusion. I'm not seeing how Nemo's proposal helps here, but I might be
missing something.
Peter
--
Peter Saint-Andre
https://andyet.com/
_______________________________________________
precis mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/precis
