John C Klensin | 17 May 2012 00:11

POP3 Spec and mail drop/ Mail store organization

These are more or less nits but some of them are a fairly large
ones (and the IMAP spec should be checked to be sure it doesn't
suffer from any of the same problems).

The second paragraph of Section 3.1 starts "Maildrops can
natively store UTF-8 or be limited to ASCII.".  Essentially the
same statement appears at the beginning of paragraph 4 only it
says "mail store" rather than "maildrop".

Issues:

(1) I'm not quite sure what the difference is between a
"maildrop" and a "mail store".  If there isn't one, the
terminology should be harmonized.  FWIW, RFC 1939 mentions
"maildrop" in several places but never mentions a "mail store"
(or "mailstore").  If there is a difference, the document should
probably go to a little more effort to identify it.  

(2) The second paragraph of Section 3.1 starts "Maildrops can
natively store UTF-8 or be limited to ASCII."  As the inclusive
statement that appears to be, it is flat wrong.  First, if the
maildrop uses Unicode, we don't care what is stored "natively",
with various flavors of UTF-16 or UTF-32 being likely (and
common).  We don't even care if messages are stored in whatever
Charset arrives over the wire (and, indeed, storage of messages
in the form in which they were delivered may be common).  We
require that headers be either ASCII or UTF-8 on the wire, but
that says almost nothing about either maildrop storage form or
message content.  What we care about is that, if the messages
are delivered across a POP3 interfaces with UTF-8 enabled, the
(Continue reading)

Barry Leiba | 17 May 2012 04:03
Picon
Favicon
Gravatar

Re: POP3 Spec and mail drop/ Mail store organization

> (1) I'm not quite sure what the difference is between a
> "maildrop" and a "mail store".  If there isn't one, the
> terminology should be harmonized.  FWIW, RFC 1939 mentions
> "maildrop" in several places but never mentions a "mail store"
> (or "mailstore").  If there is a difference, the document should
> probably go to a little more effort to identify it.

I think there is no difference.  I think 1939 used "maildrop" because
the model was the real post-office model: mail gets dropped there, and
the POP client picks it up (and it no longer resides at the drop, but
is now at the client).  With UIDL and the proliferation of "leave on
server" features, the maildrop has in many use cases given way to the
mail store model, where the server provides long-term storage for the
messages, even after they're picked up by (copied to, in this model)
the client.

I think we should stick with the terminology used in 1939.[1]

Barry

[1] Said with full realization that the extent to which that looks
like a year might be telling us that we're trying to live in the past.
Jiankang YAO | 25 May 2012 11:32
Picon

Re: POP3 Spec and mail drop/ Mail store organization


----- Original Message ----- 
From: "John C Klensin" <klensin <at> jck.com>
To: <ima <at> ietf.org>
Sent: Thursday, May 17, 2012 6:11 AM
Subject: [EAI] POP3 Spec and mail drop/ Mail store organization

> These are more or less nits but some of them are a fairly large
> ones (and the IMAP spec should be checked to be sure it doesn't
> suffer from any of the same problems).
> 
> The second paragraph of Section 3.1 starts "Maildrops can
> natively store UTF-8 or be limited to ASCII.".  Essentially the
> same statement appears at the beginning of paragraph 4 only it
> says "mail store" rather than "maildrop".
> 
> Issues:
> 
> (1) I'm not quite sure what the difference is between a
> "maildrop" and a "mail store".  If there isn't one, the
> terminology should be harmonized.  FWIW, RFC 1939 mentions
> "maildrop" in several places but never mentions a "mail store"
> (or "mailstore").  If there is a difference, the document should
> probably go to a little more effort to identify it.  
> 

yes, it should stick to maildrop.

> (2) The second paragraph of Section 3.1 starts "Maildrops can
> natively store UTF-8 or be limited to ASCII."  As the inclusive
(Continue reading)

John C Klensin | 27 May 2012 16:13

Re: POP3 Spec and mail drop/ Mail store organization


--On Friday, May 25, 2012 17:32 +0800 Jiankang YAO
<yaojk <at> cnnic.cn> wrote:

> 
> ----- Original Message ----- 
> From: "John C Klensin" <klensin <at> jck.com>
> To: <ima <at> ietf.org>
> Sent: Thursday, May 17, 2012 6:11 AM
> Subject: [EAI] POP3 Spec and mail drop/ Mail store organization
>...

>> (2) The second paragraph of Section 3.1 starts "Maildrops can
>> natively store UTF-8 or be limited to ASCII."  As the
>> inclusive statement that appears to be, it is flat wrong.
>> First, if the maildrop uses Unicode, we don't care what is
>> stored "natively", with various flavors of UTF-16 or UTF-32
>> being likely (and common).  We don't even care if messages
>> are stored in whatever Charset arrives over the wire (and,
>> indeed, storage of messages in the form in which they were
>> delivered may be common).  We require that headers be either
>> ASCII or UTF-8 on the wire, but that says almost nothing
>> about either maildrop storage form or message content.  What
>> we care about is that, if the messages are delivered across a
>> POP3 interfaces with UTF-8 enabled, the header material is in
>> UTF-8 on the wire and on delivery to the POP client.
>> 
> 
> From my point of view, I see no problem of specifying
> "Maildrops can  natively store UTF-8 or be limited to ASCII."
(Continue reading)


Gmane