Re: [imapext] AD review of draft-ietf-imapapnd-appendlimit-extension-06 (Section 2)

[Alexey Melnikov <[e-mail]>]

Dave,

On 11/12/2015 09:20, Dave Cridland wrote:
> On 10 December 2015 at 17:19, Alexey Melnikov 
> <[e-mail] <mailto:[e-mail]>> wrote:
>
>     On 10/12/2015 16:32, S Moonesamy wrote:
>
>         Hi Naren,
>         At 07:52 10-12-2015, Naren wrote:
>
>             Our MAY means
>             (a) The IMAP server can advertise this capability before
>             the user has logged in.
>             OR
>             (b) The IMAP server can advertise this capability after
>             the user has logged in.
>
>
>         Based on the AD review [1] and the short discussion about that
>         sentence in Section 2, my opinion is that the intent of the
>         what is written in the draft is not clear.  What are the
>         implications of (a) and (b)? What if the IMAP server
>         advertises that there is a limit which applies to all
>         mailboxes before the user has logged in?
>
>     The it means that the limit applies to all users. However the
>     server can also return another limit for some users after logging
>     in, which will override the globally announced limit. So it is not
>     OR, but AND/OR.
>
>
> I feel this is much more confusing than it should be.
>
> If the server gives a pre-auth APPENDLIMIT, but does not give a 
> post-auth APPENDLIMIT,

To make sure that we are all on the same page, I wrote down rules for 
CAPABILITY responses/response codes. I am sure it is incomplete and 
might be even wrong in some respects, but it should be a start:

<https://datatracker.ietf.org/doc/draft-melnikov-imap-capabilities/?include_text=1>

> does this mean that there is a general append limit but it does not 
> apply to me, or that there is a general append limit and there's no 
> change for me,

Taking into consideration the above, the "no change for me" case.

> or is it an error?

Let's talk about CAPABILITY response codes only (automatically sent by 
the server) and ignore CAPABILITY commands.

<client connects to the server>
S: * OK [CAPABILITY IMAP4rev1 LITERAL+ APPENDLIMIT=4096 STARTTLS 
AUTH=SCRAM-SHA-1] Blah-blah server ready
C: t1 STARTTLS
S: t1 OK
<TLS handshake>
C: t2 AUTHENTICATE ...
  ...
S: t2 OK [CAPABILITY IMAP4rev1 LITERAL+ APPENDLIMIT=8192 STARTTLS 
AUTH=SCRAM-SHA-1] Authenticated

So, the last value always wins. Also, once the server is returned 
APPENDLIMIT, it has to include it in all subsequent CAPABILITY response 
codes. So the server can't just omit the second APPENLIMIT.