0

For my current project I want to implement or (better re-use) a

Bounded chunked ringbuffer for packets of variable size.

Each reader should maintain his own read pointer; that means each reader should read all packets (if reading fast enough). If a writer catches up with a reader the reader pointer should be moved forward to the next valid packet.

While looking for existing works and papers I found all kind of Lock-free Queues, FIFOs etc. but none of these implements the above mentioned semantics!

Does one know about an existing implementation that implements this kind of semantics or possibly can be adapted to support it?

Helpful pointers are highly welcome!

fhw72
  • 1,066
  • 1
  • 11
  • 19
  • 1
    Sounds like LMAX Disruptor http://lmax-exchange.github.com/disruptor/. SO discussion http://stackoverflow.com/questions/6559308/how-does-lmaxs-disruptor-pattern-work .There is a c++ implementation somewhere as well –  Dec 07 '12 at 11:09
  • Looks quite interesting. Unfortunately the C++ port is really outdated but I'll take a deeper look! Thanks! – fhw72 Dec 07 '12 at 13:11

0 Answers0