james-server-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Serge Knystautas <ser...@lokitech.com>
Subject Re: Local vs Remote delivery failures
Date Fri, 06 Sep 2002 15:51:01 GMT
Yeah, I like that idea of a bounce processor for the RemoteDelivery.

So the thing is, the LocalDelivery mailet DOESN'T handle a bounce 
condition.  The process that calles LocalDelivery is supposed to check 
that this is a valid mailbox before calling LocalDelivery to store it... 
the code you are looking at in LocalDelivery (that's sending it to the 
error processor) should only get used if the mail repository is corrupt 
or there's some other configuration error.  You use a matcher before 
calling LocalDelivery so that you can then handle delivering to default 
mailboxes or a custom bounce or what-have-you.

RemoteDelivery also has a few states and additional info because it will 
contain who it was trying to connect to and the error messages. 
RemoteDelivery also has retries in certain cases, you actually have a 
variety of "bounce"-esque conditions... I've seen some servers give a 
warning notice when a message doesn't deliver immediately and gets put 
in the try-again-later bucket.

Anyway, just some thoughts... please feel free to improve on it however 
you think best.  I'm still too busy besides lurking and moderating. :(

Serge Knystautas
Loki Technologies - Unstoppable Websites

Danny Angus wrote:
> worthwhile, yes.
> Although the mailet API has bounce() in the mailet context I'm pretty sure
> that creating a bounce processor which would allow people to configure
> bounces would be worthwhile. I think you'd have to alter all the bounce()
> methods to send their mail to the bounce processor though.
>>-----Original Message-----
>>From: Noel J. Bergman [mailto:noel@devtech.com]
>>Sent: 06 September 2002 15:46
>>To: James Developers List
>>Subject: RE: Local vs Remote delivery failures
>>>>Why isn't the handling consistent?
>>>probably because this is an Open Source project, and any number of
>>>people have any number of different ideas.
>>That was my first thought, but checking the code and CVS first,
>>it appeared
>>to come from the same origin.  So I thought that there might be
>>some deeper
>>reason for why it was handled two different ways; a reason I had missed.
>>Similarly with my question about instrumenting LocalDelivery and
>>RemoteDelivery to accept a processor name for failure notification.  I
>>wasn't making the assumption that it hadn't been considered in the past.
>>Instead I was asking.
>>Should I take your responses to mean that you don't know of any
>>reasons for
>>the difference, and that you believe it might be worthwhile (post-2.1?) to
>>make such a change?
>>	--- Noel

To unsubscribe, e-mail:   <mailto:james-dev-unsubscribe@jakarta.apache.org>
For additional commands, e-mail: <mailto:james-dev-help@jakarta.apache.org>

View raw message