4 SPEC - Spec not finalized
20 - Put a consensus in place of the empty fallback-consensus file in
21 src/config and see what breaks.
22 - let bridges set relaybandwidthrate as low as 5kb
23 - we need a config option to turn off proposal 109 behavior, else
24 running a private tor network on your own computer is very hard.
25 - Make BEGIN_DIR mandatory for asking questions of bridge authorities?
26 (but only for bridge descriptors. not for ordinary cache stuff.)
27 - write a tor-gencert man page
29 Features blocking 0.2.0.x:
30 - mirror tor downloads on (via) tor dir caches
33 . geoip caching and publishing for bridges
36 o Code to load a geoip file from disk
40 o Code to store a GEOIP file in memory.
41 o Code to remember client IPs.
42 o Code to generate history lines
43 o Controller interface
44 - Track consecutive time up, not time since last-forgotten IP.
47 - Run a tweaked version on peacetime. Verify lack of kaboom.
48 - Write some basic unit tests.
49 - Mention in dir-spec.txt
50 - Mention in control-spec.txt
51 - we count hits from our reachability tests too. stop that somehow.
52 - have normal relays report geoip stats too.
53 - different thresholds for bridges than for normal relays.
54 d let Vidalia use the geoip data too rather than doing its own
56 R - bridge communities
59 - man page entries for Alternate*Authority config options
60 o make the Alternate*Authority config options pass a "default_type"
61 to parse_dir_server_line() so they don't demand as much redundancy
62 o some sort of reachability testing on bridges
63 o a dir-based way to fetch the bridge networkstatus list
64 o there's a config option with a password and somehow use an
65 http header and check it
66 - clients who have a password configured decide to ask their bridge
67 authority for a networkstatus
68 - be able to have bridges that aren't in your torrc
69 d interface for letting soat modify flags that authorities assign
71 S * tor usb windows image (vidalia, polipo, tor, firefox)
72 S/M o vidalia can launch firefox
73 - build a community version of firefox
74 . pick our favorite extensions
76 Things we'd like to do in 0.2.0.x:
77 - if we notice a cached-status directory and we're not serving v2 dir
78 info and it's old enough, delete it. same with cached-routers*.
79 - document the "3/4 and 7/8" business in the clients fetching consensus
80 documents timeline. then document the bridge user download timeline.
81 o refactor circuit_launch_by_foo so all those flags at the end are
82 actually a flags argument.
83 - config option __ControllerLimit that hangs up if there are a limit
84 of controller connections already.
85 - See also Flyspray tasks.
86 - See also all items marked XXXX020 and DOCDOC in the code
87 - https://www.torproject.org/eff/legal-faq.html#License doesn't mention
88 licenses for other components of the bundles.
90 - Before the feature freeze: (Nick)
91 D 118 if feasible and obvious
92 D Maintain a skew estimate and use ftime consistently.
93 - 105+TLS, if possible.
95 - New list of ciphers for clients
96 o Servers detect new ciphers, and only send ID cert when they
97 get an older cipher list, and only request client cert when
98 they get an older cipher list.
99 - Clients only send certificates when asked for them.
100 o Servers disable callback once negotiation is finished, so
101 that renegotiation happens according to the old rules.
102 o Clients initiate renegotiation immediately on completing
104 o Servers detect renegotiation, and if there is now a client
105 cert, they adust the client ID.
108 - New revised handshake: post-TLS:
109 - start by sending VERSIONS cells
110 - once we have a version, send a netinfo and become open
111 - Ban most cell types on a non-OPEN connection.
113 - Don't extend a circuit over a noncanonical connection with
115 - Learn our outgoing IP address from netinfo cells?
117 o Earliest stages of 110 (infinite-length) in v2 protocol:
118 add support for RELAY_EARLY.
120 - get more v3 authorities before 0.2.0.x comes out.
121 - brainstorm about who those should be
124 - Bug reports Roger has heard along the way that don't have enough
125 details/attention to solve them yet.
126 - arma noticed that when his network went away and he tried
127 a new guard node and the connect() syscall failed to it,
128 the guard wasn't being marked as down. 0.2.0.x.
129 - after being without network for 12 hours, arma's tor decided
130 it couldn't fetch any network statuses, and never tried again
131 even when the network came back and arma clicked on things.
133 . man page entry for HidServDirectoryV2 and
134 MinUptimeHidServDirectoryV2.
135 d Tor logs the libevent version on startup, for debugging purposes.
136 This is great. But it does this before configuring the logs, so
137 it only goes to stdout and is then lost.
138 d we should do another bandwidth test every 12 hours or something
139 if we're showing less than 50KB and our bandwidthrate says we can
140 do more than that. I think some servers are forgetting the results
141 of their first test, and then never seeing use.
144 o 101: Voting on the Tor Directory System (plus 103)
145 D Use if-modified-since on consensus download
147 - GETINFO to get consensus
148 - Event when new consensus arrives
149 - 105: Version negotiation for the Tor protocol
150 . 111: Prioritize local traffic over relayed.
151 - Merge into tor-spec.txt.
154 . Make cells get buffered on circuit, not on the or_conn.
155 . Switch to pool-allocation for cells?
156 - Benchmark pool-allocation vs straightforward malloc.
157 - Adjust memory allocation logic in pools to favor a little less
159 . Remove socketpair-based bridges conns, and the word "bridge". (Use
160 shared (or connected) buffers for communication, rather than sockets.)
162 - Handle rate-limiting on directory writes to linked directory
163 connections in a more sensible manner.
164 - Find more ways to test this.
165 D Do TLS connection rotation more often than "once a week" in the
167 D Streamline how we pick entry nodes: Make choose_random_entry() have
168 less magic and less control logic.
169 - Refactor networkstatus generation:
170 - Include "v" line in getinfo values.
173 . Bridges users (rudimentary version)
174 o Ability to specify bridges manually
175 o Config option 'UseBridges' that bridge users can turn on.
176 o uses bridges as first hop rather than entry guards.
177 o if you don't have any routerinfos for your bridges, or you don't
178 like the ones you have, ask a new bridge for its server/authority.
179 . Ask all directory questions to bridge via BEGIN_DIR.
180 - use the bridges for dir fetches even when our dirport is open.
181 R - drop 'authority' queries if they're to our own identity key; accept
183 X Design/implement the "local-status" or something like it, from the
184 "Descriptor purposes: how to tell them apart" section of
185 http://archives.seul.org/or/dev/May-2007/msg00008.html
186 o timeout and retry schedules for fetching bridge descriptors
187 - give extend_info_t a router_purpose again
188 o react faster to download networkstatuses after the first bridge
190 o be more robust to bridges being marked as down and leaving us
191 stranded without any known "running" bridges.
193 - Features (other than bridges):
194 - Audit how much RAM we're using for buffers and cell pools; try to
196 - Base relative control socket paths on datadir.
197 - Make TrackHostExits expire TrackHostExitsExpire seconds after their
198 *last* use, not their *first* use.
199 D Limit to 2 dir, 2 OR, N SOCKS connections per IP.
200 - Or maybe close connections from same IP when we get a lot from one.
201 - Or maybe block IPs that connect too many times at once.
204 N - Hack up a client that gives out weird/no certificates, so we can
205 test to make sure that this doesn't cause servers to crash.
208 - can we deprecate 'getinfo network-status'?
209 - can we deprecate the FastFirstHopPK config option?
213 - Quietly document NT Service options
214 - More prominently, we should have a recommended apps list.
215 - recommend pidgin (gaim is renamed)
216 - unrecommend IE because of ftp:// bug.
217 - we should add a preamble to tor-design saying it's out of date.
218 . Document transport and natdport in a good HOWTO.
219 - Publicize torel. (What else?
220 . Finish path-spec.txt
223 P - Plan a switch to polipo. Perhaps we'll offer two http proxies in
225 P - Make documentation realize that location of system configuration file
226 will depend on location of system defaults, and isn't always /etc/torrc.
227 P - Figure out why dll's compiled in mingw don't work right in WinXP.
228 - Create packages for Nokia 800, requested by Chris Soghoian
229 P - Consider creating special Tor-Polipo-Vidalia test packages,
230 requested by Dmitri Vitalev
231 o Get Vidalia supporting protocolinfo and using auth by default.
232 P - create a "make win32-bundle" for vidalia-privoxy-tor-torbutton bundle
233 P - Flyspray 487, create a universal binary privoxy for inclusion in
237 - when somebody uses the controlport as an http proxy, give them
238 a "tor isn't an http proxy" error too like we do for the socks port.
239 - we try to build 4 test circuits to break them over different
240 servers. but sometimes our entry node is the same for multiple
241 test circuits. this defeats the point.
242 - enforce a lower limit on MaxCircuitDirtiness and CircuitBuildTimeout.
243 - configurable timestamp granularity. defaults to 'seconds'.
244 - consider making 'safelogging' extend to info-level logs too.
245 - we should consider a single config option TorPrivateNetwork that
246 turns on all the config options for running a private test tor
247 network. having to keep updating all the tools, and the docs,
249 - consider whether a single Guard flag lets us distinguish between
250 "was good enough to be a guard when we picked it" and "is still
251 adequate to be used as a guard even after we've picked it". We should
252 write a real proposal for this.
253 - switch out privoxy in the bundles and replace it with polipo.
254 - make the new tls handshake blocking-resistant.
255 - figure out some way to collect feedback about what countries are using
256 bridges, in a way that doesn't screw anonymity too much.
257 - let tor dir mirrors proxy connections to the tor download site, so
258 if you know a bridge you can fetch the tor software.
259 - more strategies for distributing bridge addresses in a way that
260 doesn't rely on knowing somebody who runs a bridge for you.
261 - A way to adjust router status flags from the controller. (How do we
262 prevent the authority from clobbering them soon afterward?)
263 - Bridge authorities should do reachability testing but only on the
264 purpose==bridge descriptors they have.
265 - Clients should estimate their skew as median of skew from servers
267 - Investigate RAM use in Tor servers.
268 - Start on the WSAENOBUFS solution.
269 - Start on Windows auto-update for Tor
271 Deferred from 0.2.0.x:
273 - 113: Simplifying directory authority administration
274 - 110: prevent infinite-length circuits (phase one)
275 - 118: Listen on and advertise multiple ports:
276 - Tor should be able to have a pool of outgoing IP addresses that it is
277 able to rotate through. (maybe. Possible overlap with proposal 118.)
278 - config option to publish what ports you listen on, beyond
279 ORPort/DirPort. It should support ranges and bit prefixes (?) too.
280 (This is very similar to proposal 118.)
282 - Internal code support for ipv6:
283 o Clone ipv6 functions (inet_ntop, inet_pton) where they don't exist.
284 - Most address variables need to become tor_addr_t
285 - Teach resolving code how to handle ipv6.
286 - Teach exit policies about ipv6 (consider ipv4/ipv6 interaction!)
288 - Let controller set router flags for authority to transmit, and for
290 - add an 'exit-address' line in the descriptor for servers that exit
291 from something that isn't their published address.
292 - More work on AvoidDiskWrites?
296 - MAYBE kill stalled circuits rather than stalled connections. This is
297 possible thanks to cell queues, but we need to consider the anonymity
299 - Implement TLS shutdown properly when possible.
301 - If the client's clock is too far in the past, it will drop (or just not
302 try to get) descriptors, so it'll never build circuits.
304 - Make resolves no longer use edge_connection_t unless they are actually
305 _on_ a socks connection: have edge_connection_t and (say)
306 dns_request_t both extend an edge_stream_t, and have p_streams and
307 n_streams both be linked lists of edge_stream_t.
308 - Generate torrc.{complete|sample}.in, tor.1.in, the HTML manual, and the
309 online config documentation from a single source.
310 - Move all status info out of routerinfo into local_routerstatus. Make
311 "who can change what" in local_routerstatus explicit. Make
312 local_routerstatus (or equivalent) subsume all places to go for "what
314 - Blocking/scanning-resistance
315 - It would be potentially helpful to respond to https requests on
316 the OR port by acting like an HTTPS server.
317 - Do we want to maintain our own set of entryguards that we use as
318 next hop after the bridge? Open research question; let's say no
319 for 0.2.0 unless we learn otherwise.
320 - Some mechanism for specifying that we want to stop using a cached
323 - Detect correct version of libraries from autoconf script.
326 - See also Flyspray tasks.
327 - See also all OPEN/ACCEPTED proposals.
328 - See also all items marked XXXX and FFFF in the code.
331 - Our current approach to block attempts to use Tor as a single-hop proxy
332 is pretty lame; we should get a better one.
333 - Allow small cells and large cells on the same network?
334 - Cell buffering and resending. This will allow us to handle broken
335 circuits as long as the endpoints don't break, plus will allow
336 connection (tls session key) rotation.
337 - Implement Morphmix, so we can compare its behavior, complexity,
338 etc. But see paper breaking morphmix.
339 - Other transport. HTTP, udp, rdp, airhook, etc. May have to do our own
340 link crypto, unless we can bully DTLS into it.
341 - Need a relay teardown cell, separate from one-way ends.
342 (Pending a user who needs this)
343 - Handle half-open connections: right now we don't support all TCP
344 streams, at least according to the protocol. But we handle all that
345 we've seen in the wild.
346 (Pending a user who needs this)
350 X turn the received socks addr:port into a digest for setting .exit
351 - handle connect-dir streams that don't have a chosen_exit_name set.
352 - Have a "Faster" status flag that means it. Fast2, Fast4, Fast8?
353 - Add an option (related to AvoidDiskWrites) to disable directory
354 caching. (Is this actually a good idea??)
355 - Add d64 and fp64 along-side d and fp so people can paste status
356 entries into a url. since + is a valid base64 char, only allow one
357 at a time. Consider adding to controller as well.
358 - Some back-out mechanism for auto-approval on authorities
359 - a way of rolling back approvals to before a timestamp
360 - Consider minion-like fingerprint file/log combination.
361 - Have new people be in limbo and need to demonstrate usefulness
362 before we approve them.
365 - Standby/hotswap/redundant hidden services.
366 . Update the hidden service stuff for the new dir approach. (Much
367 of this will be superseded by 114.)
368 - switch to an ascii format, maybe sexpr?
369 - authdirservers publish blobs of them.
370 - other authdirservers fetch these blobs.
371 - hidserv people have the option of not uploading their blobs.
372 - you can insert a blob via the controller.
373 - and there's some amount of backwards compatibility.
374 - teach clients, intro points, and hidservs about auth mechanisms.
375 - come up with a few more auth mechanisms.
376 - auth mechanisms to let hidden service midpoint and responder filter
378 - Let each hidden service (or other thing) specify its own
380 - Hidserv offerers shouldn't need to define a SocksPort
383 X When we notice a 'Rejected: There is already a named server with
384 this nickname' message... or maybe instead when we see in the
385 networkstatuses that somebody else is Named with the name we
386 want: warn the user, send a STATUS_SERVER message, and fall back
388 - If the server is spewing complaints about raising your ulimit -n,
389 we should add a note about this to the server descriptor so other
390 people can notice too.
391 - When we hit a funny error from a dir request (eg 403 forbidden),
392 but tor is working and happy otherwise, and we haven't seen many
393 such errors recently, then don't warn about it.
396 - Implement missing status events and accompanying getinfos
398 - BAD_DIR_RESPONSE (Unexpected directory response; maybe we're behind
400 - BAD_PROXY (Bad http or https proxy)
401 - UNRECOGNIZED_ROUTER (a nickname we asked for is unavailable)
402 - Status events related to hibernation
403 - something about failing to parse our address?
404 from resolve_my_address() in config.c
405 - sketchy OS, sketchy threading
406 - too many onions queued: threading problems or slow CPU?
407 - Implement missing status event fields:
408 - TIMEOUT on CHECKING_REACHABILITY
409 - GETINFO status/client, status/server, status/general: There should be
410 some way to learn which status events are currently "in effect."
411 We should specify which these are, what format they appear in, and so
413 - More information in events:
414 - Include bandwidth breakdown by conn->type in BW events.
415 - Change circuit status events to give more details, like purpose,
416 whether they're internal, when they become dirty, when they become
417 too dirty for further circuits, etc.
418 - Change stream status events analogously.
419 - Expose more information via getinfo:
420 - import and export rendezvous descriptors
421 - Review all static fields for additional candidates
422 - Allow EXTENDCIRCUIT to unknown server.
423 - We need some way to adjust server status, and to tell tor not to
424 download directories/network-status, and a way to force a download.
425 - Make everything work with hidden services
427 - Performance/resources
428 - per-conn write buckets
429 - separate config options for read vs write limiting
430 (It's hard to support read > write, since we need better
431 congestion control to avoid overfull buffers there. So,
432 defer the whole thing.)
433 - Look into pulling serverdescs off buffers as they arrive.
434 - Rate limit exit connections to a given destination -- this helps
435 us play nice with websites when Tor users want to crawl them; it
436 also introduces DoS opportunities.
437 - Consider truncating rather than destroying failed circuits,
438 in order to save the effort of restarting. There are security
439 issues here that need thinking, though.
440 - Handle full buffers without totally borking
441 - Rate-limit OR and directory connections overall and per-IP and
445 - Hold-open-until-flushed now works by accident; it should work by
447 - Display the reasons in 'destroy' and 'truncated' cells under
449 - Make router_is_general_exit() a bit smarter once we're sure what
451 - Automatically determine what ports are reachable and start using
452 those, if circuits aren't working and it's a pattern we
453 recognize ("port 443 worked once and port 9001 keeps not
457 - some better fix for bug #516?
458 - don't do dns hijacking tests if we're reject *:* exit policy?
459 (deferred until 0.1.1.x is less common)
462 - Servers might check certs for known-good ssl websites, and if
463 they come back self-signed, declare themselves to be
464 non-exits. Similar to how we test for broken/evil dns now.
465 - Authorities should try using exits for http to connect to some
466 URLS (specified in a configuration file, so as not to make the
467 List Of Things Not To Censor completely obvious) and ask them
468 for results. Exits that don't give good answers should have
469 the BadExit flag set.
470 - Alternatively, authorities should be able to import opinions
471 from Snakes on a Tor.
472 - More consistent error checking in router_parse_entry_from_string().
473 I can say "banana" as my bandwidthcapacity, and it won't even squeak.
474 - Bind to random port when making outgoing connections to Tor servers,
475 to reduce remote sniping attacks.
476 - Audit everything to make sure rend and intro points are just as
477 likely to be us as not.
478 - Do something to prevent spurious EXTEND cells from making
479 middleman nodes connect all over. Rate-limit failed
480 connections, perhaps?
481 - DoS protection: TLS puzzles, public key ops, bandwidth exhaustion.
484 - Now that we're avoiding exits when picking non-exit positions,
485 we need to consider how to pick nodes for internal circuits. If
486 we avoid exits for all positions, we skew the load balancing. If
487 we accept exits for all positions, we leak whether it's an
488 internal circuit at every step. If we accept exits only at the
489 last hop, we reintroduce Lasse's attacks from the Oakland paper.
491 - Windows server usability
492 - Solve the ENOBUFS problem.
493 - make tor's use of openssl operate on buffers rather than sockets,
494 so we can make use of libevent's buffer paradigm once it has one.
495 - make tor's use of libevent tolerate either the socket or the
496 buffer paradigm; includes unifying the functions in connect.c.
497 - We need a getrlimit equivalent on Windows so we can reserve some
498 file descriptors for saving files, etc. Otherwise we'll trigger
499 asserts when we're out of file descriptors and crash.
500 - Merge code from Urz into libevent
501 - Make Tor use evbuffers.
504 - a way to generate the website diagrams from source, so we can
505 translate them as utf-8 text rather than with gimp. (svg? or
507 . Flesh out options_description array in src/or/config.c
508 . multiple sample torrc files
509 . figure out how to make nt service stuff work?
511 - Refactor tor man page to divide generally useful options from
513 - Add a doxygen style checker to make check-spaces so nick doesn't drift
514 too far from arma's undocumented styleguide. Also, document that
515 styleguide in HACKING. (See r9634 for example.)
516 - exactly one space at beginning and at end of comments, except i
517 guess when there's line-length pressure.
518 - if we refer to a function name, put a () after it.
519 - only write <b>foo</b> when foo is an argument to this function.
520 - doxygen comments must always end in some form of punctuation.
521 - capitalize the first sentence in the doxygen comment, except
523 - avoid spelling errors and incorrect comments. ;)
526 - The Debian package now uses --verify-config when (re)starting,
527 to distinguish configuration errors from other errors. Perhaps
528 the RPM and other startup scripts should too?
529 - add a "default.action" file to the tor/vidalia bundle so we can
530 fix the https thing in the default configuration:
531 http://wiki.noreply.org/noreply/TheOnionRouter/TorFAQ#PrivoxyWeirdSSLPort
534 - Patch privoxy and socks protocol to pass strings to the browser.
537 Documentation, non-version-specific.
539 - Mark up spec; note unclear points about servers
540 NR - write a spec appendix for 'being nice with tor'
541 - Specify the keys and key rotation schedules and stuff
542 - Mention controller libs someplace.
543 - Remove need for HACKING file.
544 - document http://wiki.noreply.org/noreply/TheOnionRouter/TransparentProxy on freebsd and osx
545 P - figure out why x86_64 won't build rpms from tor.spec
546 P - figure out spec files for bundles of vidalia-tor-polipo
547 P - figure out polipo install scripts for bundles of vidalia-tor-polipo on osx, win32
548 - figure out selinux policy for tor
549 P - change packaging system to more automated and specific for each
550 platform, suggested by Paul Wouter
551 P - Setup repos for redhat and suse rpms & start signing the rpms the
552 way package management apps prefer
555 J - tor-in-the-media page
556 P - Figure out licenses for website material.
557 (Phobos reccomends the Open Publication License at
558 http://opencontent.org/openpub/)
559 P - put the logo on the website, in source form, so people can put it on
560 stickers directly, etc.
561 P - put the source image for the stickers on the website, so people can
563 P - figure out a license for the logos and docs we publish (trademark
565 (Phobos reccomends the Open Publication License at
566 http://opencontent.org/openpub/)
567 R - make a page with the hidden service diagrams.
568 P - ask Jan/Jens to be the translation coordinator? add to volunteer page.
569 - add a page for localizing all tor's components.
570 - It would be neat if we had a single place that described _all_ the
571 tor-related tools you can use, and what they give you, and how well they
572 work. Right now, we don't give a lot of guidance wrt
573 torbutton/foxproxy/privoxy/polipo in any consistent place.
574 P - create a 'blog badge' for tor fans to link to and feature on their
575 blogs. A sample can be found at http://interloper.org/tmp/tor/tor-button.png
578 - make a mailing list with the mirror operators
579 - make an automated tool to check /project/trace/ at mirrors to
580 learn which ones are lagging behind.
581 - auto (or manually) cull the mirrors that are broken; and
582 contact their operator?
583 - a set of instructions for mirror operators to make their apaches
584 serve our charsets correctly, and bonus points for language
586 - figure out how to load-balance the downloads across mirrors?
587 - ponder how to get users to learn that they should google for
588 "tor mirrors" if the main site is blocked.
589 - find a mirror volunteer to coordinate all of this