On Thu, 13 Dec 2001, Sam Varshavchik wrote:
> Soon thereafter it was determined that there were some technical issues with
> that approach, and the correct way was to use the message's internal
> tracking id, and blow the message away using the cancel mechanism, hence the
> control file parameter was replaced with the message queue id parameter.
>
> The sample dupfilter was subsequently written. dupfilter can cancel dupes
> that have not been delivered yet, using their message queue id, that it gets
> from courierfilter, so your patch is going to break it.
I think the original behavior (handing off control files) is the more
flexible solution. The queue id, if required, can be read from the
control files. The control files, under the current implimentation, can
not be derived from the message queue id.
The control files, as you point out in your previous message to Jim, are
necessary to do things like filter based on the recipient, or the sender,
or the remote relay.
It seems that the simplest way to cancel a message that hasn't been
delivered is to return an SMTP style error code, as the documentation
suggests. Do you disagree?
> I think that probably the best thing to do is to append the control filename
> to the message queue id, instead of replacing it (and modify dupfilter
> accordingly). The message queue id itseld cannot contain any spaces, so
> appending a space, and the control filename, should work.
That seems overly complex. Why not feed the filter the data file name,
the message queue id, and then the control files, if necessary? Then
filters that expect the queue id (are there any other than the samples?)
will get it, and filters that want the control files will get them. At
worst, the filters that read the message id first will believe that it's
an inaccessible control file, which *may* not be a fatal error.
--
If I had a dollar for every brain that you don't have,
I'd have one dollar. - Squidward to SpongeBob
_______________________________________________
courier-users mailing list
[EMAIL PROTECTED]
Unsubscribe: https://lists.sourceforge.net/lists/listinfo/courier-users