4 SPEC - Spec not finalized
16 N - Mark up spec; note unclear points about servers
17 N - Mention controller libs someplace.
18 D FAQ entry: why gnutls is bad/not good for tor
19 P - flesh out the rest of the section 6 of the faq
20 P - gather pointers to livecd distros that include tor
21 R . more pictures from ren. he wants to describe the tor handshake, i want to
22 talk about hidden services.
23 NR- write a spec appendix for 'being nice with tor'
24 - tor-in-the-media page
25 - Remove need for HACKING file.
28 - we need to merge documentation and support
30 - and merge developers into documentation too
31 - or at least, remove developers from the title bar
32 - and remove home and make the "Tor" picture be the link to home.
33 - put the logo on the website, in source form, so people can put it on
34 stickers directly, etc.
37 N - look at the proposed os x uninstaller:
38 http://archives.seul.org/or/talk/Jan-2006/msg00038.html
39 - support dir 503s better
40 - clients don't log as loudly when they receive them
41 - they don't count toward the 3-strikes rule
42 - should there be some threshold of 503's after which we give up?
43 - think about how to split "router is down" from "dirport shouldn't
44 be tried for a while"?
45 - authorities should *never* 503 a cache, but *should* 503 clients
46 when they feel like it.
47 - update dir-spec with what we decided for each of these
48 N - commit edmanm's win32 makefile to tor cvs contrib
49 N - use only the three dirservers for hidden service descriptors
50 - and publish hidden service descriptors less frequently
51 N - when we're running an unrecommended version, we should be told
52 what versions *are* recommended. whose list do we use?
53 N o add a GUARD flag to the network-status entries.
55 - add TestSocks config option to man page
56 R - MaxConn is still in man page. needs to be ConnLimit and needs change.
57 - when logging unknown http headers, this could include bad escape codes?
58 - more generally, attacker-controller log entries with newlines in them
59 are dangerous for our users.
60 - add os x uninstaller from or-talk.
61 - make log entries include function names in win32 again.
62 - Mysterious crash for fast servers: Bug 234.
63 - Make "setconf" and "hup" behavior cleaner for LINELIST config
64 options (e.g. Log). Bug 238.
65 - Were we going to load unrecognized 'state' variables into some
66 list somewhere, and write them out whenever we update the state?
67 To be forwards and backwards compatible.
68 R - our round-robin reading is not very fair when we have few tokens
69 left in our bucket. be more fair?
70 R - streamline how we define a guard node as 'up'. document it
72 R - reduce log severity for guard nodes.
73 R - make guard node timeout higher.
74 R - We should do address rewriting after the controller sends us an
75 "attachstream 0": Bug 225.
76 R - failed rend desc fetches sometimes don't get retried.
77 R - Add config options to not publish and not fetch rend descs.
78 - Add controller interfaces to hear rend desc events and learn
79 about rend descs. In base16 I guess for now.
81 N - In circuit_get_open_circ_or_launch(),
82 /* XXXX011 NM This should be a generic "retry all directory fetches". */
84 N - if they're trying to be a tor server and they're running
85 win 98 or win me, tell them that they'll likely crash.
87 N - Display the reasons in 'destroy' and 'truncated' cells under some
90 R - Christian Grothoff's attack of infinite-length circuit.
91 the solution is to have a separate 'extend-data' cell type
92 which is used for the first N data cells, and only
93 extend-data cells can be extend requests.
94 - Specify, including thought about
97 R - When we connect to a Tor server, it sends back a signed cell listing
98 the IP it believes it is using. Use this to block dvorak's attack.
99 Also, this is a fine time to say what time you think it is.
100 - Verify that a new cell type is okay with deployed codebase
104 - find 10 dirservers.
105 - Make it no longer default for v2 dirservers to support v1.
106 - non-versioning dirservers don't need to set recommended*versions.
107 - non-naming dirservers don't need to have an approved-routers file.
108 - What are criteria to be a dirserver? Write a policy.
114 Deferred from 0.1.1.x:
116 - <weasel> it would be nice to support a unix socket for the control thing.
118 - the tor client can do the "automatic proxy config url" thing?
120 R - clients prefer to avoid exit nodes for non-exit path positions.
122 - Automatically determine what ports are reachable and start using
123 those, if circuits aren't working and it's a pattern we recognize
124 ("port 443 worked once and port 9001 keeps not working").
126 N - Should router info have a pointer to routerstatus?
127 - We should at least do something about the duplicated fields.
129 N . Additional controller features
130 - change circuit status events to give more details, like purpose,
131 whether they're internal, when they become dirty, when they become
132 too dirty for further circuits, etc.
133 R - What do we want here, exactly?
134 N - Specify and implement it.
135 - Change stream status events analogously.
136 R - What do we want here, exactly?
137 N - Specify and implement it.
138 - Make other events "better".
139 - Change stream status events analogously.
140 R - What do we want here, exactly?
141 N - Specify and implement it.
142 - Make other events "better" analogously
143 R - What do we want here, exactly?
144 N - Specify and implement it.
145 . Expose more information via getinfo:
146 - import and export rendezvous descriptors
147 - Review all static fields for additional candidates
148 - Allow EXTENDCIRCUIT to unknown server.
149 - We need some way to adjust server status, and to tell tor not to
150 download directories/network-status, and a way to force a download.
151 - It would be nice to request address lookups from the controller
153 - Make everything work with hidden services
155 X switch accountingmax to count total in+out, not either in or
156 out. it's easy to move in this direction (not risky), but hard to
157 back out if we decide we prefer it the way it already is. hm.
160 - see if we should make use of truncate to retry
161 R - kill dns workers more slowly
164 . Some back-out mechanism for auto-approval
165 - a way of rolling back approvals to before a timestamp
166 - Consider minion-like fingerprint file/log combination.
168 - config option to publish what ports you listen on, beyond
169 ORPort/DirPort. It should support ranges and bit prefixes (?) too.
171 - Relay this in networkstatus.
173 - Non-directories don't need to keep descriptors in memory.
174 o Make descriptor-fetching happen via an indirection function.
175 - Remember file and offset.
176 - Keep a journal FD for appending router descriptors.
178 - packaging and ui stuff:
179 . multiple sample torrc files
182 . figure out how to make nt service stuff work?
184 . Add version number to directory.
185 N - Vet all pending installer patches
186 - Win32 installer plus privoxy, sockscap/freecap, etc.
187 - Vet win32 systray helper code
190 - torcp needs more attention in the tor-doc-win32.
192 - unrecommend IE because of ftp:// bug.
193 - torrc.complete.in needs attention?
195 - Bind to random port when making outgoing connections to Tor servers,
196 to reduce remote sniping attacks.
197 - Have new people be in limbo and need to demonstrate usefulness
198 before we approve them.
199 - Clients should estimate their skew as median of skew from servers
202 - Alices avoid duplicate class C nodes.
203 - Analyze how bad the partitioning is or isn't.
205 . Update the hidden service stuff for the new dir approach.
206 - switch to an ascii format, maybe sexpr?
207 - authdirservers publish blobs of them.
208 - other authdirservers fetch these blobs.
209 - hidserv people have the option of not uploading their blobs.
210 - you can insert a blob via the controller.
211 - and there's some amount of backwards compatibility.
212 - teach clients, intro points, and hidservs about auth mechanisms.
213 - come up with a few more auth mechanisms.
215 . Come up with a coherent strategy for bandwidth buckets and TLS. (The
216 logic for reading from TLS sockets is likely to overrun the bandwidth
217 buckets under heavy load. (Really, the logic was never right in the
218 first place.) Also, we should audit all users of get_pending_bytes().)
219 - Make it harder to circumvent bandwidth caps: look at number of bytes
220 sent across sockets, not number sent inside TLS stream.
222 - Make router_is_general_exit() a bit smarter once we're sure what it's for.
224 - rewrite how libevent does select() on win32 so it's not so very slow.
225 - Write limiting; separate token bucket for write
226 - Audit everything to make sure rend and intro points are just as likely to
228 - Do something to prevent spurious EXTEND cells from making middleman
229 nodes connect all over. Rate-limit failed connections, perhaps?
232 Major items for 0.1.2.x:
236 N - Better hidden service performance
239 - Better estimates in the directory of whether servers have good uptime
240 (high expected time to failure) or good guard qualities (high
242 - memory usage on dir servers.
244 N - oprofile including kernel time.
246 Topics to think about during 0.1.2.x development:
247 - Figure out non-clique.
248 - Figure out partial network knowledge.
249 - Figure out incentives.
253 - Limit to 2 dir, 2 OR, N SOCKS connections per IP.
254 - Handle full buffers without totally borking
255 - Rate-limit OR and directory connections overall and per-IP and
257 - Hold-open-until-flushed now works by accident; it should work by
259 - DoS protection: TLS puzzles, public key ops, bandwidth exhaustion.
261 - tor-resolve script should use socks5 to get better error messages.
262 - make min uptime a function of the available choices (say, choose 60th
263 percentile, not 1 day.)
264 - Track uptime as %-of-time-up, as well as time-since-last-down.
265 - hidserv offerers shouldn't need to define a SocksPort
266 * figure out what breaks for this, and do it.
267 - auth mechanisms to let hidden service midpoint and responder filter
269 - Relax clique assumptions.
270 X start handling server descriptors without a socksport?
271 - tor should be able to have a pool of outgoing IP addresses
272 that it is able to rotate through. (maybe)
275 - Patch privoxy and socks protocol to pass strings to the browser.
276 - Standby/hotswap/redundant hidden services.
277 - Robust decentralized storage for hidden service descriptors.
278 - The "China problem"
279 - Allow small cells and large cells on the same network?
280 - Cell buffering and resending. This will allow us to handle broken
281 circuits as long as the endpoints don't break, plus will allow
282 connection (tls session key) rotation.
283 - Implement Morphmix, so we can compare its behavior, complexity, etc.
284 - Other transport. HTTP, udp, rdp, airhook, etc. May have to do our own
285 link crypto, unless we can bully openssl into it.
286 - Need a relay teardown cell, separate from one-way ends.
287 (Pending a user who needs this)
288 - Handle half-open connections: right now we don't support all TCP
289 streams, at least according to the protocol. But we handle all that
290 we've seen in the wild.
291 (Pending a user who needs this)