1 PHASE #1: (Goal: settle key design questions)
4 - test fails with fresh /tmp directory (but passes when run a second time)
5 problem seems to be caused by HELLO validation (unvalidated
6 HELLO not used to connect for good, then somehow SETKEY never happens);
7 I suspect the code simply drops messages that happen while no validated
8 HELLO is available; this maybe OK, but I need to find the spot and
9 add at least an INFO-log message; also should then fix testcase to retry.
12 * improve disk API [Nils] (Nils, is this done? -Christian)
13 * Windows: use events instead of pipes to signal select()s [Nils]
14 * only connect() sockets that are ready (select()) [Nils]
15 [On W32, we need to select after calling socket before
19 For these functions, it would be nice if we had testcases ("make check")
20 that would cause them to be executed and check that they are working:
21 * gnunet-service-peerinfo:
22 - change_host_trust / flush_trust
24 - discard_hosts_helper / cron_clean_data_hosts
25 * gnunet-service-transport:
26 - try_unvalidated_addresses
27 - lookup_address_callback
28 - lookup_hello_callback
29 - plugin_env_lookup_address
30 - notify_clients_disconnect
31 - list_validated_addresses
33 - disconnect_neighbour
35 * plugin_transport_tcp.c:
37 - tcp_plugin_address_pretty_printer / append_port
38 - tcp_plugin_set_receive_quota
41 - GNUNET_TRANSPORT_set_qutoa / send_set_quota
45 - remove_from_any_list / remove_neighbour
46 - GNUNET_TRANSPORT_notify_transmit_ready_cancel
49 - solicit_traffic / copy_and_free
50 - GNUNET_CORE_peer_configure / produce_configure_message
51 * gnunet-service-core:
54 - handle_client_request_configure
57 - handle_transport_notify_disconnect
58 * hostlist (everything)
59 * topology (everything)
63 PHASE #2: (Goal: recover basic file-sharing functionality)
66 * implement sqlite-based sqstore/datastore service
67 + implement service (datastore + talks to plugin)
68 + implement library (talks to service)
69 + implement sqlite plugin (talks to DB)
70 + fix testcases (make them use CPS, complete their inner workings...)
71 * implement sqlite-based dstore services
73 + implement library (talks to service)
74 + implement service (talks to plugin)
75 + implement sqlite plugin (talks to DB)
83 PHASE #3: (Goal: ready for pre-release) [completion-goal: end of 2009]
93 * UDP backend (need LIBRARY to support (de)fragmentation!)
96 + Testcases for set_quota, timeouts, disconnects, transmit_ready_cancel
97 + Better coverage of gnunet-service-transport (hello validation)
98 + direct test of plugins compliance to plugin API
101 * sqlite, mysql & postgres backend
107 - crypto_hash: use libgcrypt (supports SHA-512 since 2003)
108 - container_bloomfilter: improve efficiency (see FIXME)
110 - inefficient memmove
112 - transport_api: support forcing disconnects through low quotas!
113 (required for working F2F support!)
114 - API: consider having core provide deadline information for each message
115 (likely important for DV plugin which wants to loop back!)
116 - implement transport API to pretty-print transport address
117 + transport_api extension (API extension!)
118 + service-transport extension (protocol extension)
119 - add calls to statistics in various places
120 - implement gnunet-transport (transport configurator / tester)
121 - UPnP-based IP detection
122 (Note: build library always, build service when libxml2/etc. are available)
123 - instantly filter addresses from *other* peers that
124 are *equal* to our own address + port (i.e., localhost:2086). We
125 no longer filter those for outgoing (helps with loopback testing
126 and keeps the code clean), but we should filter strictly *impossible*
127 incoming addresses! This is for efficiency, not correctness.
128 - We currently are happy to take any address told to us in a WELCOME
129 to our set of addresses; we should have some minimal threshold-based
130 scheme, limiting both the total number of addresses that we accept
131 this way as well as requiring multiple confirmations; also, we
132 should possibly try to confirm that the given address works for
133 us ourselves (loopback-style) before adding it to the list
135 + we may be able to simplify WELCOME messages (no need to add
136 addresses there anymore, but may help to learn them there anyway...).
137 + we probably want some kind of voting/counting for learning IP addresses
138 (maybe including IP addresses in ads proportional to how often others
139 report them? we at least need some protection against >64k HELLOs!),
140 + provide a way to give the user a list of "learned" IP addresses and
141 a way to easily "veto" addresses off the list!
142 => If MiM attacker uses vetoed address, blacklist the specific IP for
143 the presumed neighbour!
144 - not sure current way of doing ACKs works well-enough
145 with unreliable transports where the ACK maybe lost;
146 the "is_new" check would then possibly prevent future
147 ACKs to be delivered, all while we're happily
148 receiving messages from that peer! Worse, the other
149 peer won't generate another ACK since it thinks we're
150 connected just fine...
152 + How necessary is ACKing in the first place? (alternatives?)
153 + Should we transmit ACKs in response to every HELLO? (would that
154 fully address the problem?)
155 - latency measurements implemented in the transport
156 plugins makes it only work for bi-di transports
157 and results in code replication
158 - should latency be included in the ReceiveCallback and
159 NotifyConnect or passed on request?
161 - auto-generate "defaults.conf" using gnunet-setup from "config.scm"
162 - integrate all options into "config.scm"
163 - change config-file writing to exclude options set to default values
165 - implement exponential back-off for service restarts
166 - better tracking of which config changes actually need to cause process restarts by ARM.
167 - have way to specify dependencies between services (to manage ARM restarts better)
168 - client-API is inefficient since it opens a TCP connection per service that is started
169 (instead of re-using connections).
171 - code currently notifies clients about "encrypted" connections being up well before
172 we get the encrypted PONG; sometimes this may be OK (for topology killing
173 unwanted connnections), but of course not in general. I suspect we want
174 to signal on PONG and have topology hook directly into transport to
175 kill plaintext connections before they have a chance to become encrypted
176 (may require minor hack in transport API)
178 - have gnunet-peerinfo print actual host addresses again
179 - add option to gnunet-peerinfo to modify trust value
181 - finish postgres implementation; simplify other SQLs using new stats
186 - active reply route caching design & implementation of service,
189 - implement advertising of hostlist URL
190 - implement learning of hostlist URLs
196 PHASE #4: [completion-goal: mid 2010]
204 * Features eliminated from util:
205 - threading (goal: good riddance!)
206 - complex logging features [ectx-passing, target-kinds] (goal: good riddance!)
207 - complex configuration features [defaults, notifications] (goal: good riddance!)
208 - network traffic monitors (goal: eliminate)
209 - IPC semaphores (goal: d-bus? / eliminate?)
211 - DNS lookup (goal: have async service; issue: still need synchronous resolution in places, current code may not be portable)
212 => code shrunk from 61 files to 34, 22k LOC to 15k LOC, 470k to 330k (with symbols)
213 * New features in util:
215 - service and program boot-strap code
216 * Major changes in util:
217 - more expressive server (replaces selector)
219 - how to integrate scheduler with GTK event loop!