Skip to content

Latest commit

 

History

History
179 lines (113 loc) · 4.7 KB

feedback.md

File metadata and controls

179 lines (113 loc) · 4.7 KB
Raw

9th NDN Hackathon Feedback & Future Work

[Project 2]: NFD on OpenWrt Home Router

Feedback:

  • Not mentioned.

Future work:

  1. Create config page for ndn-autoconfig section.
  2. Add dynamic updates for face status page.
  3. Integrate ndn6-file-server in UCl and LuCl.
  4. Integrate ndncert client in LuCl
  5. Publish as opkg package(s).

[Project 5]: Simulating Multicast Supression Scheme

Feedback:

  • Not mentioned.

Future work:

  • Not mentioned.

[Project 7]: Service Discovery for IoT Devices through PSync FullSync

Feedback:

(Problems with current PSync implementation for IoT)

  1. It doesn't efficiently support an application to be a full consumer only - one who doesn't produce any data.
  2. Harder to deploy PSync in low constraints IoT devices (consumers only) because of its intensive IBF related computation (compression, hashing, insertion ...) - mostly not needed if no data is produced,

Future work:

  1. Complete the porting of full consumer in ndn-lite and release the code
  2. Integrate in esp8266ndn library as it support ndn-lite
  3. Develop a sync protocol adaptor to support various sync protocol (similar to what NLSR is doing right now)

[Project 8]: NDN Mailing List Search Tool

Feedback:

  • Not mentioned

Future work:

  • Right now, the project only support searching in ndn-interest mailing list, in the future, the project plans to support more NDN mailing lists.

[Project 9]: NFD-Android Enhancements

Feedback:

  • Lack of API to retrive strategy choice for a given prefix

Future work:

  • Uploading the NFD-Android ping server patch.

[Project 10]: Enhanced Testbed Monitoring and FCH Service

Feedback:

  1. Send control command to /localhop/nfd/rib/register
  2. "Origin" field of control parameters MUST contain a magic number (65) for client-to-nlsr readvertisement to work
    • Not documented anywhere!
  3. Even with the origin value, the control command was still being rejected by the remote NFD
    • Impossible to debug
    • RIB management protocol does not document potential error codes, or anything else regarding the format of error messages
    • PyNDN does not provide a reason in the prefix registration failure callback
    • NFD knows the root cause (in this case a slightly out-of-sync clock) but only encodes a generic "command rejected" in the response packet
  4. Debugging
    • SSH access to testbed router (LIP6)
    • Enable debug loggign (which module? Hint: not what you'd think)
    • Command interest validator is in ndn-cxx (how to discover logger module name?)
    • Need to group logging by logical function performed, not by C++ class that implements it
    • "Timestamp is outside the grace period for key /foo/bar/..."

Future work:

  • Not mentioned.

[Project 11]: NDN-Lite over LoRa Network

Feedback:

  • Not mentioned

Future work:

  • Not mentioned.

[Project 12]: Migrate multipath forwarding strategy to latest ndnSIM and integrate ns3 TrafficControlLayer

Feedback:

  • Not mentioned.

Future work:

  • Not mentioned

[Project 13]: NDNCERT-v2: Debug, Deploy, Dominate

Feedback:

Crashed

  • Ndncert-client
    • Type enter on step 0 -> Crashed: stoi
    • Crash when entering any other string by exact challenge name
      • What would you expect to enter to the question
        • Step 3: Please type in the challenge ID from the following challenges
        • Email
  • NDNCert Server
    • Crash when receiving Interest with incorrect sha256-params
    • Crash on invalid input
      • wrong request, incorrect encryption, missing blocks in parameters TLV, anything non-conformant.

Critical Unimplemented Things

  • Hkdf derived key not used
    • In both client and server code, m_aesKey being initialized and derived using hkdf function, but not actually used anywhere
  • PROBE not implemented
    • Probe function is neither fully documented (as a spec) nor implemented anywhere in the server code. The code only inlcude functionality to set a "handler", but no uses of that (except the test cases).
    • Effectively, it is not hard-coded to return randomly generated number as identity.

Ohter (Unexpected) Things

  • Loca-ndncert-anchor not used
    • m_localNdncertAnchor (local-ndncert-anchor)
  • Client tool asks for email twice (?!?)
  • Why json for NEW/and other exchanges requires "\n" at the end ?!?
    • Not having it crashes the server
  • Minor: JSON_CA_EQUEST_ID -> typo

Future work: (Roadblocks to Domination)

  • Problems ralted to encryption/decryption
  • Stuck with decrypting JSON:
    • IV, payload, algorithm seem to be the same outputted in Python and received in C++, but C++ crashed with a mystery error from security::transform::blockCipher.