1 PHASE #1: (Goal: settle key design questions)
4 * API review: should "bpm" and "last_activity" arguments be
5 included in the ClientEventHandler?
7 * Internal: bandwidth allocation (inbound limits, trust)
8 * Internal: advertising (propagate other peers' HELLOs, find new ones)
9 * Internal: bootstrapping
10 => bootstrap should use plugins, possible multiple at the same time!
13 * Should "server" argument be given in event callbacks?
14 * consider adding "get_time" to "configuration" API
17 PHASE #2: (Goal: recover basic core functionality)
20 * implement sqlite-based sqstore/datastore service
21 + implement library (talks to service)
22 + implement service (datastore + talks to plugin)
23 + implement sqlite plugin (talks to DB)
24 * implement sqlite-based dstore services
25 + implement library (talks to service)
26 + implement service (talks to plugin)
27 + implement sqlite plugin (talks to DB)
30 * implement hostlist service (need to bootstrap!)
35 * mysql & postgres backend
38 * UDP backend (need LIBRARY to support (de)fragmentation!)
41 + Testcases for set_quota, timeouts, disconnects, transmit_ready_cancel
42 + Better coverage of gnunet-service-transport (hello validation)
43 + direct test of plugins compliance to plugin API
46 PHASE #3: (Goal: ready for pre-release) [completion-goal: end of 2009]
61 - transport_api: support forcing disconnects through low quotas!
62 - API: consider having core provide priority and possibly
63 deadline information for each message
64 (likely important for DV plugin which wants to loop back!)
65 - implement transport API to pretty-print transport address
66 + transport_api extension (API extension!)
67 + service-transport extension (protocol extension)
68 - add calls to statistics in various places
69 - implement gnunet-transport (transport configurator / tester)
70 - UPnP-based IP detection
71 (Note: build library always, build service when libxml2/etc. are available)
72 - instantly filter addresses from *other* peers that
73 are *equal* to our own address + port (i.e., localhost:2086). We
74 no longer filter those for outgoing (helps with loopback testing
75 and keeps the code clean), but we should filter strictly *impossible*
76 incoming addresses! This is for efficiency, not correctness.
77 - We currently are happy to take any address told to us in a WELCOME
78 to our set of addresses; we should have some minimal threshold-based
79 scheme, limiting both the total number of addresses that we accept
80 this way as well as requiring multiple confirmations; also, we
81 should possibly try to confirm that the given address works for
82 us ourselves (loopback-style) before adding it to the list
84 - not sure current way of doing ACKs works well-enough
85 with unreliable transports where the ACK maybe lost;
86 the "is_new" check would then possibly prevent future
87 ACKs to be delivered, all while we're happily
88 receiving messages from that peer! Worse, the other
89 peer won't generate another ACK since it thinks we're
90 connected just fine...
92 + How necessary is ACKing in the first place? (alternatives?)
93 + Should we transmit ACKs in response to every HELLO? (would that
94 fully address the problem?)
95 - latency measurements implemented in the transport
96 plugins makes it only work for bi-di transports
97 and results in code replication
98 - should latency be included in the ReceiveCallback and
99 NotifyConnect or passed on request?
101 - auto-generate "defaults.conf" using gnunet-setup from "config.scm"
102 - integrate all options into "config.scm"
103 - change config-file writing to exclude options set to default values
105 - implement exponential back-off for service restarts
106 - better tracking of which config changes actually need to cause process restarts by ARM.
107 - have way to specify dependencies between services (to manage ARM restarts better)
109 - have gnunet-peerinfo print actual host addresses again
110 - add option to gnunet-peerinfo to modify trust value
112 - finish postgres implementation; simplify other SQLs using new stats
117 - active reply route caching design & implementation of service,
122 PHASE #4: [completion-goal: mid 2010]
130 * Features eliminated from util:
131 - threading (goal: good riddance!)
132 - complex logging features [ectx-passing, target-kinds] (goal: good riddance!)
133 - complex configuration features [defaults, notifications] (goal: good riddance!)
134 - network traffic monitors (goal: eliminate)
135 - IPC semaphores (goal: d-bus? / eliminate?)
137 - DNS lookup (goal: have async service; issue: still need synchronous resolution in places, current code may not be portable)
138 => code shrunk from 61 files to 34, 22k LOC to 15k LOC, 470k to 330k (with symbols)
139 * New features in util:
141 - service and program boot-strap code
142 * Major changes in util:
143 - more expressive server (replaces selector)
145 - how to integrate scheduler with GTK event loop!
150 * UTIL: 75%, 4914 out of 6463