POP3 highest number accessed

[John Gardiner Myers <[e-mail]>]

Mark Crispin <[e-mail]> writes:
>      In RFC-1225, the ``highest number accessed'' could be initially set to
> some value other than zero when a user authenticates.  This would happen if it
> was known that the message had already been accessed (in c-client/IMAP terms,
> the \Seen flag was set).  The RSET command restores this ``highest number
> accessed'' to the value at initial startup.
> 
>      In RFC-1460, there is no change (that I can determine) to the ability of
> the initial ``highest number accessed'' to be other than zero with certain
> servers, but the RSET command has been changed to set this to zero.
> 
>      John Myers tells me that in draft-rose-again-01.txt, the ``highest number
> accessed'' is supposed to be set to zero.  This document claims that ``no one
> implemented'' preserving that state across sessions; this is incorrect since I
> have implemented it.
> 
>      My implementation is based upon RFC-1081, and checked against RFC-1225.
> I am having trouble understand what I am supposed to do.  If the feature of
> preserving the ``highest number accessed'' across sessions was nuked because
> ``no one implemented'' it, that reason is incorrect.  My server, which has
> been widely distributed, has implemented it since it was written in November
> 1990.
> 
>      I am seriously confused as to what, if anything, I should do.  It is not
> going to be possible for me to recall earlier versions of my POP3 server; like
> it or not, it is out there and being used.  Instead, let me turn to specific
> questions:
>  1) Am I obligated to delete support for a capability that I implemented in
>     good faith, and who knows what might depend upon it now?
>  2) Which POP3 document do I believe?
>  3) If I support the concept of preserving the ``highest number accessed''
>     across sessions (as exists in all the RFCs; only draft-rose-again-01.txt
>     changes this), what does it mean to follow the dictate of RFC-1460 that
>     the RSET command sets the ``highest number accessed'' to zero?  Am I
>     supposed to erase the seen state of all messages, so that the next session
>     will have a ``highest number accessed'' of zero?  [I object to this,
>     because it will screw up non-POP3 c-client access, e.g. Pine, IMAP, etc.
>     But it's the only interpretation that makes sense if I obey RFC-1460 and
>     have state preservation across sessions.]
>  4) If the ``highest number accessed'' is not to be preserved across sessions,
>     what do I tell people who might complain that I broke it in a new version
>     of my POP3 server?  Remember, I have no idea who may or may not depend
>     upon my server preserving this data.  Granted, that preserving this data
>     was always clearly optional, but now it seems to be banned.

Mark Crispin <[e-mail]> writes:
> I feel a need to clarify my position on POP3
 [...]
> I do not wish to argue any particular side on the questions I have raised
> (even if it sounds as if I have).  What I want is a ruling from the POP3
> community on the issue, so I can pass the buck on any bad problems.  If, for
> example, I'm told that I must not preserve status, I will happily tell anyone
> who complains to me ``blame the POP3 guys, I told them about it and they said
> to do it that way.''

-- 
_.John G. Myers		Internet: [e-mail]
			LoseNet:  ...!seismo!ihnp4!wiscvm.wisc.edu!give!up