Re: [idn] ietf london idn wg meeting minutes

[Dan Oscarsson <Dan.Oscarsson@trab.se>]

>> >ACE EVALUATION WITH IDNs ALREADY REGISTERED, Yoshiro Yoneya
>> >
>> >- Most important evaluation criterion to study is to maximize number
>> >   of characters, raw speed is less important because nameprep is the
>> >   slow stage.
>>
>> This is not correct in general. At my site which uses the Latin-1
>subset
>> of UCS, most ACE encodings are a very expensive step compared to
>nameprep.
>
>I am curious about this statement. Do you have any numbers for this?
>
>You see, all datas from other sources, from Verisign, i-DNS.net, JPNIC
>etc who have done Nameprep-ACE/UTF-8 have clearly shown that Nameprep ~=
>10xACE in terms of CPU time.

The nameprep step for Latin-1 is as simple as for ASCII. Even larger sets
of UCS where simple alphabets are used will have a very simple and
quick nameprep step.
For Latin-1 it is like for ASCII - it can be done with a simple table lookup.
To do ACE with compression includes much more work - just look at some of
the algorithms.

Note that I said subset. At my site I do not need to support more
than Latin-1. host names outside that scope will remain ACE. And if I implement
this support in my DNS server, it will also only need to support
Latin-1 when comparing names as I will only have latin-1-names in
my authorative domains.

   Dan