Re: SMTP and unicode

From: Hans Aberg (haberg@math.su.se)
Date: Wed May 18 2005 - 08:29:00 CDT

  • Next message: Alexander Kh.: "Re: Corrections to Glagolitic"

    At 14:40 +0200 2005/05/18, Antoine Leca wrote:
    >On Wednesday, May 18, 2005 2:24 AM Hans Aberg wrote:
    >
    >> The simplest way would be to write the SMTP protocol to require 8-bit
    >> uses. Then those mail servers still out there which zeroes the 8'th
    >> bit are no longer SMTP compliant, and must be changed. People have
    >> now had more than ten years to change their mail server software, so
    >> it should be not a big problem.
    >
    >RFC 2821 (which indeed asks servers to support the 8BITMIME extension, using
    >SHOULD) is "only" 4 year old (RFC 1652 did not require server support.)
    >Also the renewing of softwares in this area is _very_ slow (why changing
    >something that just works?)
    >And some mail administrators are quite angry when you ask them to update
    >their softwares, but more importantly the vast majority of them quite simply
    >do not attend your requests (because of the excess of junk mail that is
    >floating around.)

    Perhaps one needs some kind of crawler program that locates the 7-bit
    only mail servers. Then one can systematically weed them out.

    >I guess RFC 2821 did the first step, which was to include the SHOULD. Its
    >revision will then put MUST (thus requiring effective upgrades),

    Right.

    >and it
    >would be the next revision after that which could allow a client (MUA or
    >MTA) to send 8-bit material without proper workaround if the server is not
    >able to deal with it... Only then you could say the castrating servers are
    >eliminated.

    Castrating the castrating mail servers. Tying it back to Unicode, it
    would seem a simplification if UTF-8 data could be passed via mail
    without squeezing it into 7-bit encodings.

    -- 
       Hans Aberg
    


    This archive was generated by hypermail 2.1.5 : Wed May 18 2005 - 08:29:42 CDT