On Wednesday, 16 בNovember 2005 22:13, Andreas Fink wrote:
> I'm -1 for this patch and -100 for that SMSC.

hear hear!

> There's no reason for an SMSC even to look at those informations. The
> SMSC does have no need to reassemble the parts of that SMS so it
> should just blindly transport every individual part. 

I agree that the SMSC should not do anything to messages being 
transported, but I don't see any reason to preclude the SMSC from 
reassembling the messages offline (out of scope, and out of the 
delivery path, w/o introducing the modified messages back into the 
path). It might be useful for traffic analysis or billing, and as long 
as the SMSC is delivering SMs according to the specs, and observing 
privacy and anonymity requirements, I don't care what they do with my 
messages afterwards.

That being said, its a given that SMSCs, regardless of what offline 
analysis they are doing, should definitely not reject perfectly valid 
messages whose only crime was specifying SAR sequence numbers out of 
order.

> You should shout 
> at the guy running this braindamaged SMSC (is it maybe from
> comverse?)

We did, and it was no help at all. It was an SMSC for a Kazachstan 
service provider, surprisingly enough it wasn't delivered by Comverse 
but by a local software company whose name currently escapes me.

I still think that there might be a problem with the kannel priority 
queue (as evident by the need for a fix - I myself have not seen this 
happening, but I never tracked the order of outgoing SAR messages) - 
SAR messages positioned on the queue should have the same priority 
parameters and therefor should be sent in the order they were 
introduced.

From looking at the code, at first I couldn't see why this should 
happen, as split messages are indeed handled in the order they are 
created everywhere I looked. The only problem may be the uuid 
generation where in a 3 part SAR message, the uuid for parts would be 
2,3,1 for some reason, but the priority queue doesn't check the SM's id 
when sorting. 

I didn't really understand the prioqueue sorting algorithm (though it 
looks like some kind of a heap tree), but it is definitely rearranges 
items with the same rank, and I think it shouldn't.

-- 
Oded Arbel
m-Wise mobile solutions
[EMAIL PROTECTED]

+972-9-9611212 (204)
+972-54-7340014

::..
Save the Earth - kill a lawyer 

Reply via email to