From: David B. <da...@ra...> - 2011-11-20 17:59:25
|
Gloria - The addition of rest octets to GSM was a bit of a hack to the standard and complicated the connection between L3 and L2. Without rest octets, the "L2 pseudolength" element on L3 RR messages sent on the CCCH exactly matched where L2 length indicator would have been placed in LAPDm header. This allowed a simplification in the pre-2.8 code, allowing us to completely ignore the "L2 pseudolength" as a concept and just generate a normal length field in L2 instead of treating CCCH messages as a special case in L3. Adding rest octets forces us to treat the L2 pseudolength as a special case now for those messages (CCCH messages) that have rest octets. So now we have the "L2 body length" that does not include rest octets and a "full body length" that does include rest octets and a new "...RO" subclass to deal with those special cases and still produce the correct frame lengths in L2. The old "bodyLength" method was removed entirely to prevent confusion, but functionally it was the same as the new "l2 body length" is now. I hope that explanation helps. -- David On Nov 20, 2011, at 7:32 AM, Gloria Mazzi wrote: > Hi all, > > considering the gap between 2.6 and 2.8, i decided to revise all whole sources to better understand the features added. > > Actually i'm following the modifications about rest-octets and i'm having quite difficulties by following all changes using diff manually. > ( Eg. besides the newly code added, i noticed that "bodyLength()" has been widely replaced with "l2BodyLength(), etc... ) > > Could i ask to David or Kurtis if it could be available the Revision Logs that were made during the development stage between 2.6 & 2.8 (or at least about rest-octets)? > > Thank you for attention. > > Cheers, > Gloria > David A. Burgess Range Networks, Inc. 560 Brannan St. San Francisco, CA 94107 USA cell +1 707 208 2622 |