Re: [idn] UTF-8 / RACE

[Keith Moore <moore@cs.utk.edu>]

> In contrast, ACE guarantees that the data is unreadable for users of
> today's non-ACE software. 

so what?  UTF-8 guarantees that the data is unreadable for users of
today's non UTF-8 software.  and even the software that today can
display UTF-8 and accept it as input, cannot reliably compare two 
UTF-8 representations of an IDN and see if they are equivalent.

> When you claim that an ACE IDN is transmitted
> perfectly, you are ignoring the final step: getting the IDN into the
> user's head!

with ACE, we have the case that zero existing software can display them
properly or accept them as input.  this software will have to be upgraded
for those who wish to use IDNs, but at least it will be obvious that this
software has faults that need to be corrected.  in the meantime, uses of
IDNs that do not require  input or display (say, replying to an email 
message, relaying mail from one MTA to another, or clicking on a URL 
link) will continue to work.

with UTF-8, we have the case that some existing software can display them
properly and/or accept them as input.  however, this software will not
nameprep IDNs accepted as input, and thus, for some languages, the 
software will not reliably lookup IDNs even though it will accept them as
input.  these faults will not be so obvious to the user as those with ACE.

summary: in either case software has to be upgraded.  ACE is less likely
to disrupt operations; its flaws are more likely to be noticed and therefore
more likely to be fixed, and ACE is more amenable to incremental upgrade.

> Yes, there are a few pieces of software that have to be upgraded and
> redeployed to make UTF-8 IDNs work. But there are many more pieces of
> software that have to be upgraded and redeployed to make ACE IDNs work.
> See http://cr.yp.to/proto/idn.html for details.

this is a mischaracterization, based on numerous erroneous assumptions.

Keith