Noel,

I tested the experimental patch sent yesterday, and it works fine.

I think that it can be useful, and it simply has no effect if the configuration 
entries are not used.

If it is accepted and added to James, I would modify the AttachmentFileNameIs and 
HasAttachment patches that I submitted in the last two days, leaving out the global 
catches. I would also prepare the patches to v3 and resubmit.

Any thoughts?

Vincenzo

> -----Original Message-----
> From: Vincenzo Gianferrari Pini
> [mailto:[EMAIL PROTECTED]
> Sent: luned� 9 giugno 2003 18.46
> To: James Developers List
> Subject: [no virus] RE: Separating internal errors from addressing
> errors in config.xml
> 
> 
> Oops, a small mistake. Better this one.
> 
> Vincenzo
> 
> > -----Original Message-----
> > From: Vincenzo Gianferrari Pini
> > [mailto:[EMAIL PROTECTED]
> > Sent: luned� 9 giugno 2003 18.31
> > To: James Developers List
> > Subject: [no virus] RE: Separating internal errors from addressing
> > errors in config.xml
> > 
> > 
> > Noel,
> > 
> > I'm sending here attached a possible patch to LinearProcessor v2 
> > that would do what I'm talking about.
> > It has not been tested, it's just as a basis of discussion, and 
> > as a way for me to get deeper in my understanding of the James 
> > code. If all this discussion dies, no problem: it has been a good 
> > experiment.
> > 
> > It would handle the following init parameter syntax:
> > 
> >     
> > <onMatchException>match|nomatch|error|aProcessorName</onMatchException>
> >     <onMailetException>ignore|error|aProcessorName</onMailetException>
> > 
> > the default being "error".
> > 
> > There is also a little change to JamesSpoolManager: it is a 
> > partially redundant mail.setState(Mail.ERROR) that I moved to 
> > another place. Partially redundant in the current code because if 
> > it gets there it is either because the requested processor was 
> > not found or because processor.service(mail) returned an 
> > exception, and in that case the state was already set to 
> > Mail.ERROR by handleException.
> > Instead with the changes to LinearProcessor the next state must 
> > be set before throwing the exception.
> > 
> > A possible problem is that an exception could occur after the 
> > mail object has been modified in place, and no "rollback" is 
> > done. But this problem occurs also in the current code to mails 
> > sent to the error processor.
> > 
> > Does it make any sense at all?
> > 
> > Vincenzo
> > 
> 


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to