Re: delivery reports and EMI [known bug issue]

On Mittwoch, November 13, 2002, at 10:53  Uhr, Stipe Tolj wrote:

>> That would not be good enough in most cases.
>
> why? I think a primary key consisting of 8 digits of the receiver
> number and the timestamp is a pretty unique thing in the DLR table?!

The point is that in some areas 8 digits would interfear already on the 
areacode level.
Its better than doing nothing though but its still not really unique.

What happens if you deliver to a short ID which is 3-4 digits only?

> Ok, if you push several messages in the same second to the same
> number, then you clash up again to the same situation where you can't
> distinguish the messages for the inbound dlr.

I agree on this but a delivery of the first message makes it very 
likely that the second message gets delivered too. So you would expect 
to have two identical deliveries or failures.

Our SMSC does work around this problem by simply creating longer 
timestamps which are truely unique (for up to 1000 messages in the same 
second to the same destination). However some EMI client software seems 
to have problems with that. Kannel has not. The first problem is the 
CMG SMSC which is not unique.

Well the world is not perfect.
Maybe we should simply go for 8 digits now until we find a better 
solution.
Got a patch?