Skip to content

Minutes from face to face meeting @ FU Berlin on September 27, 2013

Oleg Hahm edited this page Feb 4, 2014 · 1 revision

Minutes from the Face-to-Face developer meeting @ FU Berlin on September 27, 2013

Local participants: Lotte Steenbrink, Christian Mehlis, Oleg Hahm, Ludwig Ortmann Skype participants: Thomas Eichinger, Martin Landsmann Minute taker: Ludwig


1.) Message API Problems 2.) Merging Policy 3.) Features for the next release 4.) C++ integration 5.) Makefiles Windows compatibility 6.) Network stack memory optimization

Message API Problems

  • invalid (internal problems in Christian's application)

Merging policy

  • reminder: anyone with commit rights should merge as soon as enough people have acknowledged the pull request and the pull request has been open for reviewing $long enough
  • trivial changes (like a new header include) can be merged immediately after the first ACK

Features for the next release

  • network stack optimization
  • POSIX features
    • time.h

      requires changes in the core

      should be fully included in the release

    • pthread.h

      probably complex

      required for libcppa

    • semaphore.h

      should be trivial (about 10 lines)

      already available as part of sixlowpan

      no named semaphores, name resolution returns NULL

      should be fully included in the release

    • Martin Lenders looks at the network part, Christian at the rest
    • Martin Landsmann checks what is needed for libcppa
  • basic support for Cortex A9
  • destiny fixes
  • CoAP
  • CCN lite

C++ integration

  • currently issues with printf and float (e.g. for the sht-11 on MSB-A2)
  • encapsulate with "extern C", pull request follows
  • find yet not declared functions and add them to the header
  • try to avoid ifdefs
  • split up the pull request (wherever possible)

Makefiles Windows compatibility

  • the according pull request needs to be updated and should be merged (no further objections)

Network stack memory optimization

  • currently every abstraction layer holds its own buffer

  • the upper layers should share one buffer, possibly using Stephan Zeisbergs memory manager

  • the lower layers (driver and transceiver module) should use one buffer together

  • possible real-time improvement: driver just informs the transceiver module, which fetches the packet from the radio

  • additional: check defined stack sizes and try to minimize them

Next meeting: Since the next release is envisioned for November (due to CoAP plugtest (20.11. and SAFEST demo (25.11)) -> 28.10 - 1.11., Doodle follows

Clone this wiki locally
You can’t perform that action at this time.