RE: SAR well tested?
"Igor Ivoilov" <[email protected]>
| Newsgroups | gmane.comp.mobile.kannel.devel |
|---|---|
| Message-ID | <[email protected]> |
You' are right. NACK is not imlemented for receiption. I did not implement it yet just because I never met a problem with that and have other very urgent stuff. > -----Original Message----- > From: Marco Fassiotto [mailto:[email protected]] > Sent: Monday, September 15, 2003 8:48 PM > To: [email protected] > Subject: SAR well tested? > > > Hello, > I started to look into the kannel code a few days ago (very few) > and I am currently following the SAR implementation, finding a > few things that need clarification (beside being a newbie on > kannel I also happen to be a newbie on WAP). > However, here is the question regarding the WTP responder. > On reception of a SARed transaction, process_sar_transaction() > adds the first invoke and the following segm-invokes to the > list of received blocks using add_sar_transaction(). > When it is time to ACK/NAK a group (GTR=1), there is no check for a > complete reception of all blocks in a group, and it is always > built an ACK packet despite the fact that one of them may have been > lost. > Sending a segmented result does not seem to suffer of the same > problem as the state machine is prepared to receive a NAK and to > resent the lost blocks. > I am sure I am missing something and hope that someone can help > shed some light. > Thanks > Marco > > >