Steve Hole (steve.hole@messagingdirect.com)
Thu, 13 Dec 2001 12:11:17 -0700
On Thu, 13 Dec 2001 20:15:29 +0200 "Rubinstein, Dmitry"
<Dmitry_Rubinstein@icomverse.com> wrote:
> In fact, support for unsolicited events (especially EXPUNGE) is, from my
> expirience, the most non-trivial part of implementing an IMAP software.
> So if anything would make the implementation of IMAP available to the
> mortals is this extension. Now that I think about it, the fact that this
> feature is not present in IMAP from the beginning looks
> like an overlook more than anything else.
Oh my!
The code to do exactly this is freely available in the uwash c-client.
It is not exactly the way that we did it, but it certainly gives you the
idea. And, no disrespect intended, it isn't that hard. It's an array.
You know how big to make the array. You can certainly put a little
structure into it. You might have to worry about how big the array gets
in a segmented memory architecture. You might have to build a list of
them. The code to manage this in the c-client is about 500 lines soaking
wet. Ours is about double that, but that is because cache management is
bound with our asynchrous event notification and io completion
notification framework. That is a more advanced architecture that most
GUI applications would be well served to try to use.
Note that this stuff ran successfully in Mac v5 and Windows 3.1 on
machines that were not too far off in terms of memory constraints and
management from modern PDA devices.
In short, having done it, the argument that it is too difficult to manage
sequence numbers and UID's is simply ludicrous. There are some
extensions that would require servers to maintain more state information
for resynchronization purposes that would make synchronizing clients more
bandwidth and storage efficient -- a
"report-changes-since-last-checkpoint" type function.
Cheers.
--- Steve Hole Chief Technical Officer - Electronic Billing and Payment Systems ACI Worldwide, Messaging Direct Mailto:Steve.Hole@MessagingDirect.com Phone: 780-424-4922
This archive was generated by hypermail 2.0b3 on Thu Dec 13 2001 - 21:12:37 IST