ietf-822
[Top] [All Lists]

Re: gzip-8bit

2003-02-26 23:18:14

Dan Kohn wrote:

I may have missed something obvious, but my quick read of
RFC 1939 says that POP3 only supports RFC 822 messages.  Can someone
comment on POP3's capabilities to deal with 8BitMIME and binary content?

POP3 message transfer (header, empty line, body) is much like SMTP;
after a RETR command, the server sends a response code followed by
byte-stuffed lines, ending the transfer with a lone dot on a line.
And as such it is possible for an implementation to be nearly
binary compatible (the data ends with CRLF.CRLF, and the optional
TOP command is expected to return some number of message _lines_).
But the interesting question is whether or not some implementor
has coded in a hard line limit below 998 octets. RFC 1939 mentions
a 512 character (incl. CRLF) limit on response lines; if a careless
implementor used that for message lines, one could have a 510 octet
limit. With few exceptions, POP doesn't really concern itself with
message content, it just moves the data from point A to point B.

Can conforming IMAP4 implementations only support 7bit bodies?  If so,
what the capabilities negotiation to indicate 8BitMIME support?

IMAP can transfer 8-bit body content, or can encode binary content
for transmission between server and client.  IMAP delves
heavily into the message structure so as to be able to provide
information derived from message and MIME part header fields, etc.
and explicitly only allows standard 7-bit header content. I don't
think there's any line length limit mentioned in the IMAP RFCs;
data transmission is generally a byte count followed by the data,
so I wouldn't expect a line length limit per se -- but I suppose
it's possible that some implementation has a limit.

From a practical point of view, all of my mail goes through a couple
of POP3 transfers, one of which is actually implemented by an IMAP
server.  I have no trouble with 8-bit body content, though 8-bit
header content can be mangled (all 8-bit-set octets replaced with
ASCII 'X') by the IMAP implementation (and that's easy to patch).
And I've seen lines longer than 998 octets.  But I may be lucky;
there could be some pathological POP3 or IMAP implementations.



<Prev in Thread] Current Thread [Next in Thread>