On Mon March 14 2005 10:32, Tony Finch wrote:
Anyone who uses such software will find it difficult to diagnose problems
and will suffer from less effective spam filtering until the software is
fixed. I don't think improvements to the specification should be held back
by idiot implementations.
The specification clearly states that order is not significant and
notes that fields may be reordered in transit. Always has said that.
The problem isn't the specification; it's the unwarranted assumption
that reordering never happens.
Did you find out how widely deployed these manglers were and whether they
were usually installed as end-points or as relays?
Most are deployed at one site fewer than if they didn't rearrange
fields, failed to comply with [E]SMTP requirements, were reasonably
configurable, etc.:-) Many were fairly widely-used MTAs; others
were POP->SMTP converters for retrieving (or forwarding) POP-based
message stores (ISP-hosted) to another message store.
In a similar discussion several months ago on the ietf-822 mailing
list, Laird Breyer stated that there are a number of database-
based UA implementations (Outlook and Lotus Notes were specifically
FWIW, I ended up with a home-brew POP->SMTP converter, sendmail,
and Cyrus IMAP.