Kosma Moczek - GSoC 2010 Proposal

I'd like to improve the UMPA library in several aspects:

• Milestone one: fix critical issues.

  • Finish and merge link-layer branch.
  • Restore IPv6 support (currently broken).
  • Complete the autogeneration support for all fields (like TCP window, etc).
  • Improve the tests to cover abnormal situations instead of just the optimistic ones.

• Milestone two: add new functionalities.

  • Add a bunch of new protocols.
    Examples include 802.11Q and ICMP from lower layers and DNS and DHCP from application layer.
  • Create a command-line interface and/or language bindings (Perl and/or C).
    Without them, the library itself is largely useless for people without Python knowledge (except for PacketManupulator, which isn't scriptable).
  • Finish the PacketManipulator+UMPA integration.

• Milestone three: additional improvements.

  • Write a more complete set of examples to give the library an easier learning curve.
  • Close all existing UMPA-related tickets.
  • Polish the code and prepare for release.

• The ultimate goal:

  • By the end of GSoC, release UMPA 0.5.

Each of those milestones should take about three to four weeks, fitting nicely in the GSoC schedule. My proposal might look a bit like a random bunch of small tasks, and in fact that's what it is: polishing all the small things needed to turn a development-grade library into a friendly and ready-to use product.

There is also a non-coding aspect I'm interested in: giving UMPA a real 'kick'. Currently googling for 'packet generator' or 'packet manipulator' doesn't show UMPA anywhere near the top of the search results. Helping people find the library they need would make the code we write for UMPA much more useful.

-- Kosma