John C Klensin wrote:
rfc2821bis-03 adds new text to explicitly compare SMTP's reply
code model to that of FTP in the hope of eliminating any
possible residual confusion about issues 18 and 28. Should that
text be retained, dropped, or modified in some way?
I vote for the Drop.
As with Chris Haynes, I think the new "up-to-date definition of the
reply code model should be complete and self-sufficient" on both counts;
persistent reply codes and eliminating of the 1yz preliminary code.
Case in point: While our own FTP product experience helped carry the
similar idea to possible mix-code, 1yz- usage for SMTP as well, the
consideration for mix codes may not only be from a FTP perspective but
rather one simply based on the now declared incorrect ambiguous current
"last code" only interpretation as shown by this SMTP client developer:
This fella was also surprised to find out that the popular Exim was not
reading the reply codes based on a "last code" interpretation which he
presumed "that's how everyone operated."
Hector Santos, CTO
Santronics Software, Inc.