[horde] Doubling emails-problem in Outlook 2013

[Patrick De Zordo]

Am 20.05.2014 19:02, schrieb Michael J Rubinsky:
>
> Quoting Patrick De Zordo <patrick at spamreducer.eu>:
>
>> Am 20.05.2014 16:00, schrieb Michael J Rubinsky:
>>> WTF? This worked for me perfectly for about an hour yesterday, and 
>>> now it only works sporadically. The log shows the same WBXML being 
>>> sent (we add the new email to the destination folder and delete the 
>>> email from the source folder). If this isn't happening in a 
>>> deterministic way, there is REALLY nothing we can do. A search on 
>>> Google reveals that this IS a known issue in OL, so I'm going to 
>>> have to say "not our problem" at this point.
>>
>> Me too, so I was sure this is perfekt! But..god.. this are the badest 
>> news for this week.
>
> Ok. I have one more idea, but it is going to require a bit of surgery. 
> Right now, we send the message ids as the raw IMAP UID for the 
> message. This means the id's, by themselves are not unique across 
> mailboxes. The specs state that the field for the message ids can be a 
> string up to 64 characters in length.
>
> Further, when I originally tested this fix, I was using a test IMAP 
> server - with only a few mailboxes and messages. When I go back to 
> test this on THAT server, it works again. It's only on my production 
> server that this fails. My guess is that this is what is happening:
>
> 1. Message with a UID of 221 is moved by user from FolderOne to 
> FolderTwo.
> 2. OL (incorrectly) automatically moves the message (maybe to provide 
> quicker UI feedback?) to FolderTwo when sending the MOVEITEMS command.
> 3. When OL receives the DELETE command from the server to remove UID 
> 221 from FolderOne - it can't find it since it's already moved to 
> FolderTwo and there are probably more than one message with a UID of 
> 221 so it doesn't know what to do. I'm thinking that in this case, OL 
> expects the UID to be a *unique* identifier across all mailboxes. So, 
> regardless of what mailbox the email is currently in, OL will know 
> where it is.
>
> To do this, though, we would have to perform some magic - like maybe 
> prepending the folder's UID to the message UID before sending it and 
> stripping it on the way back. Though off hand, I think this would 
> cause problems in some commands when we don't already know the folder 
> id. The other option is to store this UID -> OL UID mapping in the 
> device state. However, this will increase the storage requirements.
>
> I'll have to think on this some more and play around when I get time.

Would it be a benefit for you having a testmachine online for testing 
out the ActiveSync Protocoll regarding DELETE and MOVE?
Could try to get a trial license and use one of our servers/virtual 
servers with a running product where you could try out a protocoll 
tracing..?

>
>
>