TODO

   1 Core
   2 ~~~~
   3 * protocols: implement dumping of protocol-dependent rte attributes
   4
   5 - io.c: refuse old-style multicasts for unnumbered interfaces?
   6
   7 - config: executable config files
   8
   9 - counters (according to SNMP MIB?)
  10 - better memory allocators
  11 - default preferences of protocols: prefer BGP over OSPF/RIP external routes?
  12
  13 - static: check validity of route destination?
  14
  15 - device: configuration of interface patterns
  16
  17 - adding of route: check whether all bits not covered by masklen are zero
  18
  19 - netlink: import Linux route attributes to our rta's, so that they can be filtered?
  20
  21 - iface: SIOCGIFINDEX exists on glibc systems, but it doesn't work on 2.0.x kernels!
  22
  23 - socket: Use IP_RECVERR for BGP TCP sockets?
  24
  25 - OSPF: refuse running on non-multicast devices
  26
  27 - config: when parsing prefix, check zero bits
  28
  29 - krt: rescan interfaces when route addition fails?
  30
  31 Cleanup
  32 ~~~~~~~
  33 - right usage of DBG vs. debug
  34 - cleanup debugging calls
  35 - logging and tracing; use appropriate log levels
  36 - check incoming packets and log errors!!
  37 - check log calls for trailing newlines
  38 - check if all protocols set proper packet priorities and TTL's.
  39 - replace all NUM, IPA and expr tokens by constant filter expressions
  40 - try compiling with -Wunused
  41 - does everybody test return value of sk_open?
  42
  43 Various ideas
  44 ~~~~~~~~~~~~~
  45 - real multipath (doesn't seem to be simple at all :()
  46 - fake multipath (even less simple)
  47 - route recalculation timing and flap dampening [see RFC2439 for algorithms]
  48 - aggregate engine: standard route aggregation and summarization
  49 - aggregate engine: injection of manually configured pseudo-static routes
  50 - generate default route if any working BGP connection exists (aggregate engine again?)
  51 - generate default route to IGP's (aggregate engine yet another time?)
  52
  53
  54 RIP
  55 ~~~
  56         - RIP: export-only and import-only mode?
  57         - drop RIPv1 (Historic protocol)?
  58         - Route Tag
  59         - limit routing table xfer (frequency, only to neighbors)
  60         - multicast on/off (per interface)
  61
  62 OSPF
  63 ~~~~
  64         - importing of device routes for networks where we don't run OSPF
  65         - check incoming packets using neighbor cache
  66         - stub area: either no external routes or only default route
  67         - RFC2328 appendix E: Use a better algorithm
  68         - automatic generation of external route tags (RFC1403)
  69         - RFC1587 NSSA areas
  70         - RFC2370 opaque LSA's
  71         - RFC1793 Demand Circuit Support ??
  72         - respect interface MTU and try not to create larger packets unless unavoidable
  73
  74
  75 BGP
  76 ~~~
  77         - in, local, out RIB
  78         - maxsize=4096
  79         - detection of loops
  80         - aggregation, ATOMIC_AGGREGATE
  81         - communities
  82         - confederations
  83         - attributes must be sorted!
  84         - re-export of NEXT_HOP attribute
  85         - BGP session over currently down interface
  86         - LOCAL_PREF attribute
  87         - error notification received -> log error
  88         - set TTL to 1 (configurable?)
  89         - consulting IGP for next-hop information? (what if it changes?)
  90         - inter-advertisement delay???!
  91         - normalize (sort) incoming AS-SET's
  92         - maximum length of AS paths
  93
  94         - expected neighbor AS
  95         - hold time
  96         - idle timer after error: initial value, exponential growth, maximum value
  97         - import of IGP routes (use external route tags from OSPF)