1 PHASE #1: (Goal: settle key design questions)
4 * Windows: use events instead of pipes to signal select()s [Nils]
5 * only connect() sockets that are ready (select()) [Nils]
6 [On W32, we need to select after calling socket before
8 * Add "DISK" API for creating of temporary files
9 (as used in datacache/ module)
11 PHASE #2: (Goal: recover basic file-sharing functionality)
16 For these functions, it would be nice if we had testcases ("make check")
17 that would cause them to be executed and check that they are working:
18 * gnunet-service-peerinfo:
19 - change_host_trust / flush_trust
21 - discard_hosts_helper / cron_clean_data_hosts
22 * gnunet-service-transport:
23 - try_unvalidated_addresses
24 - lookup_address_callback
25 - lookup_hello_callback
26 - plugin_env_lookup_address
27 - notify_clients_disconnect
28 - list_validated_addresses
30 - disconnect_neighbour
32 * plugin_transport_tcp.c:
34 - tcp_plugin_address_pretty_printer / append_port
35 - tcp_plugin_set_receive_quota
38 - GNUNET_TRANSPORT_set_qutoa / send_set_quota
42 - remove_from_any_list / remove_neighbour
43 - GNUNET_TRANSPORT_notify_transmit_ready_cancel
44 - Testcases for set_quota, timeouts, disconnects, transmit_ready_cancel
45 - gnunet-service-transport HELLO validation (how good is our coverage?)
46 - direct test of plugins compliance to plugin API
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)
62 Module features to implement:
63 * DATACACHE (needed for DHT)
64 - implement testcases for dstore API
65 - implement testcases for dstore plugin
66 - implement performance tests
67 * TESTING (needed for DV, DHT, Topology)
69 - implement library for local testing
70 - implement testcases for library
71 - implement library for distributed testing
72 - implement testcases for distributed testing
74 - implement testcases (needs TESTING)
79 - implement DV service
80 - implement DV library
81 - implement DV transport plugin
83 - implement performance tests
86 - implement DHT service (needs DV, DATACACHE)
87 - implement DHT library
89 - implement performance tests
92 - design network structs (CS)
93 - implement FS library
96 + persistence mechanism
106 - design network structs (P2P)
107 - datastore request queueing mechanism
108 - implement FS service (needs DHT)
113 - implement testcases
116 + sharing API (needs TESTING)
125 - implement performance tests (needs TESTING)
133 PHASE #3: (Goal: ready for pre-release) [completion-goal: end of 2009]
135 Module features to implement:
138 - need to settle basic design; do we want to keep guile?
140 - good to have for DV evaluation!
142 - good to have for DV/DHT evaluation!
152 Plugins to implement:
153 * UDP backend (need LIBRARY to support (de)fragmentation!)
160 Minor TODO items / known bugs:
162 - crypto_hash: use libgcrypt (supports SHA-512 since 2003)
163 - container_bloomfilter: improve efficiency (see FIXME)
165 - inefficient memmove
167 - transport_api: support forcing disconnects through low quotas!
168 (required for working F2F support!)
169 - API: consider having core provide deadline information for each message
170 (likely important for DV plugin which wants to loop back!)
171 - implement transport API to pretty-print transport address
172 + transport_api extension (API extension!)
173 + service-transport extension (protocol extension)
174 - add calls to statistics in various places
175 - implement gnunet-transport (transport configurator / tester)
176 - UPnP-based IP detection
177 (Note: build library always, build service when libxml2/etc. are available)
178 - instantly filter addresses from *other* peers that
179 are *equal* to our own address + port (i.e., localhost:2086). We
180 no longer filter those for outgoing (helps with loopback testing
181 and keeps the code clean), but we should filter strictly *impossible*
182 incoming addresses! This is for efficiency, not correctness.
183 - We currently are happy to take any address told to us in a WELCOME
184 to our set of addresses; we should have some minimal threshold-based
185 scheme, limiting both the total number of addresses that we accept
186 this way as well as requiring multiple confirmations; also, we
187 should possibly try to confirm that the given address works for
188 us ourselves (loopback-style) before adding it to the list
190 + we may be able to simplify WELCOME messages (no need to add
191 addresses there anymore, but may help to learn them there anyway...).
192 + we probably want some kind of voting/counting for learning IP addresses
193 (maybe including IP addresses in ads proportional to how often others
194 report them? we at least need some protection against >64k HELLOs!),
195 + provide a way to give the user a list of "learned" IP addresses and
196 a way to easily "veto" addresses off the list!
197 => If MiM attacker uses vetoed address, blacklist the specific IP for
198 the presumed neighbour!
199 - not sure current way of doing ACKs works well-enough
200 with unreliable transports where the ACK maybe lost;
201 the "is_new" check would then possibly prevent future
202 ACKs to be delivered, all while we're happily
203 receiving messages from that peer! Worse, the other
204 peer won't generate another ACK since it thinks we're
205 connected just fine...
207 + How necessary is ACKing in the first place? (alternatives?)
208 + Should we transmit ACKs in response to every HELLO? (would that
209 fully address the problem?)
210 - latency measurements implemented in the transport
211 plugins makes it only work for bi-di transports
212 and results in code replication
213 - should latency be included in the ReceiveCallback and
214 NotifyConnect or passed on request?
215 - FIXME's with latency being simply set to 0 in a few places
216 - Memory leak (running valgrind --trace-children=yes on test_transport_api:
217 ==28393== 16 bytes in 1 blocks are indirectly lost in loss record 1 of 5
218 ==28393== at 0x4C2260E: malloc (vg_replace_malloc.c:207)
219 ==28393== by 0x52343E3: GNUNET_xmalloc_unchecked_ (common_allocation.c:62)
220 ==28393== by 0x5234389: GNUNET_xmalloc_ (common_allocation.c:53)
221 ==28393== by 0x524458A: GNUNET_NETWORK_socket_create_from_accept (network.c:289)
222 ==28393== by 0x524B2DA: ??? (server.c:332)
223 ==28393== by 0x524A4C7: ??? (scheduler.c:425)
224 ==28393== by 0x524A73D: GNUNET_SCHEDULER_run (scheduler.c:510)
225 ==28393== by 0x524FF8C: GNUNET_SERVICE_run (service.c:1326)
226 ==28393== by 0x405500: main (gnunet-service-transport.c:2645)
228 ==28393== 65,744 (65,728 direct, 16 indirect) bytes in 1 blocks are definitely lost in loss record 5 of 5
229 ==28393== at 0x4C2260E: malloc (vg_replace_malloc.c:207)
230 ==28393== by 0x52343E3: GNUNET_xmalloc_unchecked_ (common_allocation.c:62)
231 ==28393== by 0x5234389: GNUNET_xmalloc_ (common_allocation.c:53)
232 ==28393== by 0x524473E: GNUNET_NETWORK_socket_create_from_accept (network.c:323)
233 (rest of trace identical)
239 - auto-generate "defaults.conf" using gnunet-setup from "config.scm"
240 - integrate all options into "config.scm"
241 - change config-file writing to exclude options set to default values
243 - implement exponential back-off for service restarts
244 - better tracking of which config changes actually need to cause process restarts by ARM.
245 - have way to specify dependencies between services (to manage ARM restarts better)
246 - client-API is inefficient since it opens a TCP connection per service that is started
247 (instead of re-using connections).
249 - code currently notifies clients about "encrypted" connections being up well before
250 we get the encrypted PONG; sometimes this may be OK (for topology killing
251 unwanted connnections), but of course not in general. I suspect we want
252 to signal on PONG and have topology hook directly into transport to
253 kill plaintext connections before they have a chance to become encrypted
254 (may require minor hack in transport API)
256 - have gnunet-peerinfo print actual host addresses again
257 - add option to gnunet-peerinfo to modify trust value
259 - finish postgres implementation; simplify other SQLs using new stats
264 - active reply route caching design & implementation of service,
267 - implement advertising of hostlist URL
268 - implement learning of hostlist URLs
274 PHASE #4: [completion-goal: mid 2010]
279 - update webpage documentation
281 - expand bibliography
282 - convert documentation pages to books
283 - update books (especially for developers)
284 - add content type for links and view
285 - create good Drupal theme for GNUnet
286 - make a NICE download page and figure out how to
287 enable developers to publish TGZs nicely
288 - port "contact" page; add impressum
289 - add content type for "todo" items?
294 * Features eliminated from util:
295 - threading (goal: good riddance!)
296 - complex logging features [ectx-passing, target-kinds] (goal: good riddance!)
297 - complex configuration features [defaults, notifications] (goal: good riddance!)
298 - network traffic monitors (goal: eliminate)
299 - IPC semaphores (goal: d-bus? / eliminate?)
301 - DNS lookup (goal: have async service; issue: still need synchronous resolution in places, current code may not be portable)
302 => code shrunk from 61 files to 34, 22k LOC to 15k LOC, 470k to 330k (with symbols)
303 * New features in util:
305 - service and program boot-strap code
306 * Major changes in util:
307 - more expressive server (replaces selector)
309 - how to integrate scheduler with GTK event loop!