On 1-Apr-2009, at 16:53, mouss wrote:
do you guys know that designing a protocole is at least as hard as
writing code? now, is there still anyone around who does believe in
perfect code?
protocol design is actually harder:
This is precisely why it hasn't happened, and why the mail system is,
at least partially, crippled under the volume of a type of mail that
didn't even exist when SMTP was introduced ~25 years ago. I *never*
said it would be easy, or simple.
- defining a protocol is easier when you have a single working
prototype. once you have hundreds of implementations, life gets hard
The trick is to make the standard rigid enough that any implementation
has to be 100% valid or it is 100% invalid. Learn from the errors of
HTML. It's either compliant, or it fails.
- email is useful because many people can communicate thanks to the
availability of those many non perfect yet working solutions. you can
start designing a new protocol. but at one time, you'll face the
"upgrade the universe" problem.
Yep, that is probably the largest problem, and is primarily why I
think a whole new standard that is in no way beholden to the previous
standards is the only way out of this mess.
if you think authentication, encryption, blahblahtion would solve it,
think again.
No, as I've aid, this will not solve the spam problem. It will solve
part of it though, and it will solve the part of it that I think is
the most abusive and most dangerous, the forged headers.
Spam is a social problem, and social problems can't be solved by
technical means only. technology des certainly help, to some extent.
Laws do help too. Russians killing a spammer helps too;-p ...
Heh. Maybe and RFC for how to best kill spammers then? :)
--
I gotta straighten my face This mellow-thighed chick just put my
spine out of place
