On Tue, 3 Feb 2004, Iljitsch van Beijnum wrote:
On 3-feb-04, at 12:18, Paul Robinson wrote:
Why? Why not insist that all e-mail addresses should be accompanied
with a
chinese representation, or a russian address? What is so special about
the
latin character set?
You mean apart from the facts that it's most widely understood by
people, the most widely implemented in computers, and all
internet-relevant standards that use text rely on it?
A problem with having two variations of the same email address used
together: I'm afraid this may be an in for email abusers, similar to
the way you can spread a worm using a file called worm.jpg.vbs because
the user thinks she's opening a JPEG file while the system thinks it's
a visual basic script.
The above is correct. Additionally the problem is that allowing multiple
charsets in the address and this not being visible to end-user can cause
considerable confusion because absolutly the same latin email address
would look the same in many other character sets: mail-ng(_at_)ietf(_dot_)org
could
easily have been us-ascii or it could be ISO-8859-1 or it could be
windows-1251 - all these charsets include standard latin alphabit althouth
last one includes additinal cyrilic characters. If we consider email
addresses in different charsets to all be different even if they use same
"character", then there are several particular problems:
1. User who is used to sending emails to addresses written in his native
character set might not recognize need to change to latin when sending
email to more common addresses spelled in latin and his email would
then bounce
2. Worth if ISPs such as AOL begin allowing end-users to get email addresses
tom(_at_)aol(_dot_)com where "tom" part is actually in say windows-1251
charset,
then we have a confusion as both of these tom(_at_)aol(_dot_)com would look
the
same to recepient allowing for various types of deception.
The above problems have been discussed many times over the years especially
as it related to IDN (international domain names) with possible solution
that every character should be considered by itself and all latin and
other characters that are exactly the same in writing should be "shared"
across charsets (this still leaves confusion regarding characters that are
very simular like various 'u' characters with dots and bars on top used
in german and other european languages) AND/OR that all addresses be
converted to latin in a way allowing parties to see they are really
different ones.
I'm afraid I dont' really see any good solution to above problems. The
only thing we can hope is that programmers of email client programs would
put additional emphasis (color highlighting, etc) whenever email address
in from or to or part has any characters different then default character
set normally used by the user.
---
William Leibzon
Elan Networks
william(_at_)elan(_dot_)net