procmail
[Top] [All Lists]

Re: Return an attachement...

1998-11-11 23:41:34
On 11 November 1998, Bart Schaefer <schaefer(_at_)zanshin(_dot_)com> wrote:
On Thu, 12 Nov 1998, Liviu Daia wrote:
[...]
    No.  The "name=..." parameter to "Content-Type:" is
non-standard, despite the fact that programs like Pine are still
happily using it.

Yes, I know it's non-standard.  But it appeared in a previous draft of
the MIME specs, to which many UAs were written, and is therefore still
fairly widely used/recognized.
[...]

    Can you please name the official document (RFC, IETF draft or
whatever) that introduces the "name=..." parameter to "Content-Type:"?
I've been trying to track the origins of this stuff for a while, without
much success.  IIRC, Eric Raymond from the DRUMS committee claims to
have had a similar experience.

OTOH, preaching the use of obsolete forms for the sake of backward
compatibility can be pretty harmful.

This is not a matter of "backwards compatibily."  We're not talking
about UA implementation here.  It's a matter of the sender producing
the desired effect for as many recipients as possible; maximal
interoperability.  Newer MIME interpreters will simply ignore the
name= parameter, so no harm is done.

Preaching that new UAs ought to expect and interpret the name=
parameter, or suggesting using c-type name= *in place of* c-disp
filename=, would be harmful, I agree.
[...]

    Somehow, you succeeded to get this backwards.

Scenario 1: newer MUAs ignore "name=..." when reading messages, but
keep using it when sending, to "produce the desired effect for as
many recipients as possible".  Result: people with newer MUAs are
slightly annoyed when receiving messages sent by older MUAs, but
finally learn to live with it.  If they are more than slightly annoyed,
they downgrade to an older version.  "name=..." is still with us
(ok, with out descendants...) in 2198, along with "filename=...",
and a bunch of new stuff "file_nickname=...", "file_surname=..." and
"file_whatever=...", that does essentially the same thing.

Scenario 2: newer MUAs accept "name=..." when reading messages (giving
priority to "filename=..." if that one is present), but only use
"filename=..." when sending.  Result: users with older MUAs (that don't
accept the newer form) are slightly annoyed when receiving messages sent
by newer MUAs, but finally learn to live with it.  If they are more
than slightly annoyed, they upgrade to a MUA that knows about the new
standard.  A few years later very few people are still using MUAs that
specify filenames with "name=...", and developers can finally forget
about it.

    Regards,

    Liviu Daia

-- 
Dr. Liviu Daia               e-mail:   daia(_at_)stoilow(_dot_)imar(_dot_)ro
Institute of Mathematics     web page: http://www.imar.ro/~daia
of the Romanian Academy      PGP key:  http://www.imar.ro/~daia/daia.asc