1 PHASE #1: (Goal: settle key design questions)
4 * Make sure ARM supports daemons (non-service background processes, see hostlist/topology)
7 * improve disk API [Nils]
8 * Windows: use events instead of pipes to signal select()s [Nils]
9 * only connect() sockets that are ready (select())
10 [Nils: I'm not sure what you mean by this; fresh sockets are always
11 ready for connect(), just 'write' after connect needs select AFAIK;
12 please clarify. --Christian]
15 For these functions, it would be nice if we had testcases ("make check")
16 that would cause them to be executed and check that they are working:
17 * gnunet-service-peerinfo:
18 - change_host_trust / flush_trust
20 - discard_hosts_helper / cron_clean_data_hosts
21 * gnunet-service-transport:
22 - try_unvalidated_addresses
23 - lookup_address_callback
24 - lookup_hello_callback
25 - plugin_env_lookup_address
26 - notify_clients_disconnect
27 - list_validated_addresses
29 - disconnect_neighbour
31 * plugin_transport_tcp.c:
33 - tcp_plugin_address_pretty_printer / append_port
34 - tcp_plugin_set_receive_quota
37 - GNUNET_TRANSPORT_set_qutoa / send_set_quota
41 - remove_from_any_list / remove_neighbour
42 - GNUNET_TRANSPORT_notify_transmit_ready_cancel
45 - solicit_traffic / copy_and_free
46 - GNUNET_CORE_peer_configure / produce_configure_message
47 * gnunet-service-core:
50 - handle_client_request_configure
53 - handle_transport_notify_disconnect
54 * hostlist (everything)
55 * topology (everything)
59 PHASE #2: (Goal: recover basic file-sharing functionality)
62 * implement sqlite-based sqstore/datastore service
63 + implement library (talks to service)
64 + implement service (datastore + talks to plugin)
65 + implement sqlite plugin (talks to DB)
66 * implement sqlite-based dstore services
67 + implement library (talks to service)
68 + implement service (talks to plugin)
69 + implement sqlite plugin (talks to DB)
76 * have ONE backend working
79 PHASE #3: (Goal: ready for pre-release) [completion-goal: end of 2009]
89 * UDP backend (need LIBRARY to support (de)fragmentation!)
92 + Testcases for set_quota, timeouts, disconnects, transmit_ready_cancel
93 + Better coverage of gnunet-service-transport (hello validation)
94 + direct test of plugins compliance to plugin API
97 * sqlite, mysql & postgres backend
103 - crypto_hash: use libgcrypt (supports SHA-512 since 2003)
104 - container_bloomfilter: improve efficiency (see FIXME)
106 - inefficient memmove
108 - transport_api: support forcing disconnects through low quotas!
109 (required for working F2F support!)
110 - API: consider having core provide deadline information for each message
111 (likely important for DV plugin which wants to loop back!)
112 - implement transport API to pretty-print transport address
113 + transport_api extension (API extension!)
114 + service-transport extension (protocol extension)
115 - add calls to statistics in various places
116 - implement gnunet-transport (transport configurator / tester)
117 - UPnP-based IP detection
118 (Note: build library always, build service when libxml2/etc. are available)
119 - instantly filter addresses from *other* peers that
120 are *equal* to our own address + port (i.e., localhost:2086). We
121 no longer filter those for outgoing (helps with loopback testing
122 and keeps the code clean), but we should filter strictly *impossible*
123 incoming addresses! This is for efficiency, not correctness.
124 - We currently are happy to take any address told to us in a WELCOME
125 to our set of addresses; we should have some minimal threshold-based
126 scheme, limiting both the total number of addresses that we accept
127 this way as well as requiring multiple confirmations; also, we
128 should possibly try to confirm that the given address works for
129 us ourselves (loopback-style) before adding it to the list
131 - not sure current way of doing ACKs works well-enough
132 with unreliable transports where the ACK maybe lost;
133 the "is_new" check would then possibly prevent future
134 ACKs to be delivered, all while we're happily
135 receiving messages from that peer! Worse, the other
136 peer won't generate another ACK since it thinks we're
137 connected just fine...
139 + How necessary is ACKing in the first place? (alternatives?)
140 + Should we transmit ACKs in response to every HELLO? (would that
141 fully address the problem?)
142 - latency measurements implemented in the transport
143 plugins makes it only work for bi-di transports
144 and results in code replication
145 - should latency be included in the ReceiveCallback and
146 NotifyConnect or passed on request?
148 - auto-generate "defaults.conf" using gnunet-setup from "config.scm"
149 - integrate all options into "config.scm"
150 - change config-file writing to exclude options set to default values
152 - implement exponential back-off for service restarts
153 - better tracking of which config changes actually need to cause process restarts by ARM.
154 - have way to specify dependencies between services (to manage ARM restarts better)
156 - have gnunet-peerinfo print actual host addresses again
157 - add option to gnunet-peerinfo to modify trust value
159 - finish postgres implementation; simplify other SQLs using new stats
164 - active reply route caching design & implementation of service,
167 - implement advertising of hostlist URL
168 - implement learning of hostlist URLs
174 PHASE #4: [completion-goal: mid 2010]
182 * Features eliminated from util:
183 - threading (goal: good riddance!)
184 - complex logging features [ectx-passing, target-kinds] (goal: good riddance!)
185 - complex configuration features [defaults, notifications] (goal: good riddance!)
186 - network traffic monitors (goal: eliminate)
187 - IPC semaphores (goal: d-bus? / eliminate?)
189 - DNS lookup (goal: have async service; issue: still need synchronous resolution in places, current code may not be portable)
190 => code shrunk from 61 files to 34, 22k LOC to 15k LOC, 470k to 330k (with symbols)
191 * New features in util:
193 - service and program boot-strap code
194 * Major changes in util:
195 - more expressive server (replaces selector)
197 - how to integrate scheduler with GTK event loop!