Table-building

[Erik van der Poel]

On 2/1/07, Kenneth Whistler <kenw at sybase.com> wrote:
> The important thing is that this scenario updates gracefully,
> and the worst condition is some false negatives for unupdated
> applications, rather than false *positive* matches, which
> *would* be catastrophic.

No, Ken, the problem with a false negative is that it doesn't let the
user know that they might have to update their app. MSIE7 could, in
principle, even ask the user "We have a new version that supports this
link. Would you like to upgrade?" (Assuming that the app "phones home"
once in a while to find out that a new version is available that
supports certain additional characters.)

And, the problem with a false positive, is that an unscrupulous or
error-prone registry operator might allow both the uppercase and
lowercase versions of a new character to be registered, leading to a
situation where the user of an old app is taken to one site, while the
user of a new app is taken to a different site, possibly owned by
someone else (i.e. xn--mza vs xn--nza), even though the link in the
HTML did not change (it still has the uppercase).

I suppose an unscrupulous or error-prone registry operator would
eventually be caught and the authorities might then redelegate that
TLD, but Firefox is, in principle, exposing its users to this problem
in the meantime.

Erik