serval-dna/TODO
paul.gardner.stephen f225465dfc Initial check in
git-svn-id: http://serval-dna.googlecode.com/svn/trunk/trunk/dna@2 6d01b88e-114a-4e5d-930d-818638e25ad7
2010-07-13 12:15:46 +00:00

39 lines
2.3 KiB
Plaintext

Pending:
- Parallelise large-value reads and writes
- Fix out-of-order problem for writing large values to stdout (file is fine)
- Non-mmap() mode for Mesh Potato and JFFS
- Return early if all peers have responded (what about broadcast address virtual-peers?)
- Activation flag in HLR records so that create without further action does not make a completed record.
- async mode: fork()/background after first response with file output (for pipelining with MP IVRs)
- async mode: work with broadcast requests
- async mode: work with instance=-1 requests
- SID based Peer targeting, e.g., for writeItem
- cache peer:SID mappings to allow later unicasting
- consider NACK for peers with no results to return
Done:
20100623 - testdna script now uses /bin/sh instead of /bin/csh facilitating testing on more embedded targets
20100623 - Retries for requests (already deals with SET, including multi-packet SET?)
20100623 - Multi-packet GET retry policy (okay, it is as bad as TFTP with retries, but that is probably okay for now,
at least we have the potential to pipe-line)
20100623 - Multi-packet set doesn't send multiple packets (ok to confirm each in turn)
20100623 - Output to file confirmed functional
20100623 - BATMAN peer list retrieval
20100623 - excluding previous responders from re-broadcast requests
20100611 - Added code for template based file output when reading variables
20100611 - Added code to allow user specified peers
20100601 - Space retries properly in time, and allow multiple recvfroms() per retry.
20100601 - Do not request results from peers that have already responsed.
20100601 - Ensure broadcasts do not result in duplicate identical results
20100601 - broadcast get: compulsory retry regime
20100531 - Simulated packet loss / BER for server for testing.
20100531 - "response set" structure for efficient chaining of responses, as well as making it easier to manage retries
20100529 - SET working (replace and noreplace modes)
20100529 - Get rid of unnecessary timeouts
20100529 - Regression test script
20100529 - Batch-mode responses (for instance=-1)
20100529 - Multi-instance requests now working (and with no unnecessary delays)
20100529 - Search by SID now works
20100529 - Reading all instances of a variable by SID without timeout, but actually reading them all