4 SPEC - Spec not finalized
16 Things we'd like to do in 0.2.0.x:
17 - See also Flyspray tasks.
18 - See also all items marked XXXX020 and DOCDOC in the code
19 - http://tor.eff.org/eff/legal-faq.html#License doesn't mention
20 licenses for other components of the bundles.
23 - Bug reports Roger has heard along that way that don't have enough
24 details/attention to solve them yet.
25 - tup said that when he set FetchUselessDescriptors, after
26 24 or 48 hours he wasn't fetching any descriptors at all
27 anymore. This was in 0.2.0 but worked fine in 0.1.2.
28 - arma noticed that when his network went away and he tried
29 a new guard node and the connect() syscall failed to it,
30 the guard wasn't being marked as down. 0.2.0.x.
31 - after being without network for 12 hours, arma's tor decided
32 it couldn't fetch any network statuses, and never tried again
33 even when the network came back and arma clicked on things.
35 - phobos says relaybandwidth* sometimes don't do what we expect.
36 http://interloper.org/tmp/2007-06-bw-usage.png
37 - this notion of authorities notifying servers that they're
38 unreachable is bunk -- it's leftover from the time when all
39 servers ran 24/7. now it triggers every time a server goes
40 away and then returns before the old descriptor has expired.
41 o 0.2.0.x dir authorities have stopped giving people Authority
43 - add a --quiet commandline option that suppresses logs. useful
44 for --hashed-password and maybe others.
45 - Tor logs the libevent version on startup, for debugging purposes.
46 This is great. But it does this before configuring the logs, so
47 it only goes to stdout and is then lost.
48 o Aug 30 [...] wmtbf 4:6:16
49 The wmtbf time value is missing some zero-pads.
52 . 101: Voting on the Tor Directory System (plus 103)
54 * Describe schedule in copious detail.
55 . Get authorities voting
56 o Code to manage key certificates
59 D Code to retry download.
60 o Code to generate consensus from a list of votes
61 o Detect whether votes are really all for the same period.
62 o Push/pull documents as appropriate.
63 o Pull votes and signatures if we don't get them.
64 o Cache votes and signatures on disk?
65 o Code to keep consensus docs in limbo if they don't have
66 have enough signatures.
68 o Have clients know which authorities are v3 authorities, and what
70 - While we're at it, let v3 authorities have fqdns lines.
71 - Fix all XXXX020s in vote code
72 . Validate information properly.
73 o Warn if we get a vote with different authorities than we know.
74 o Don't count votes with a different valid-after when generating
76 - Dump certificates with the wrong time. Or just warn?
77 o Warn authority ops when their certs are nearly invalid.
78 - When checking a consensus, make sure that its times are plausible.
79 o Add a function that will eventually tell us about our clock skew.
80 For now, just require that authorities not be skewed.
81 - Start caching consensus documents once authorities make them
82 - Start downloading and using consensus documents once caches serve them
84 - GETINFO to get consensus
85 - Event when new consensus arrives
86 o 104: Long and Short Router Descriptors
87 o Drop bandwidth history from router-descriptors around September 10,
88 once torstatus and weasel have upgraded.
89 - 105: Version negotiation for the Tor protocol
90 . 111: Prioritize local traffic over relayed.
92 - Merge into tor-spec.txt.
95 . Make cells get buffered on circuit, not on the or_conn.
96 . Switch to pool-allocation for cells?
97 - Benchmark pool-allocation vs straightforward malloc.
98 - Adjust memory allocation logic in pools to favor a little less
100 . Remove socketpair-based bridges conns, and the word "bridge". (Use
101 shared (or connected) buffers for communication, rather than sockets.)
103 - Handle rate-limiting on directory writes to linked directory
104 connections in a more sensible manner.
105 - Find more ways to test this.
106 - Have clients do TLS connection rotation less often than "every 10
107 minutes" in the thrashy case, and more often than "once a week" in the
109 - Streamline how we pick entry nodes: Make choose_random_entry() have
110 less magic and less control logic.
111 o Maybe move NT services into their own module.
112 - Refactor networkstatus generation:
113 - Include "v" line in getinfo values.
116 . Bridges users (rudimentary version)
117 o Ability to specify bridges manually
118 o Config option 'UseBridges' that bridge users can turn on.
119 o uses bridges as first hop rather than entry guards.
120 o if you don't have any routerinfos for your bridges, or you don't
121 like the ones you have, ask a new bridge for its server/authority.
122 . Ask all directory questions to bridge via BEGIN_DIR.
123 - use the bridges for dir fetches even when our dirport is open.
124 R - drop 'authority' queries if they're to our own identity key; accept
126 X Design/implement the "local-status" or something like it, from the
127 "Descriptor purposes: how to tell them apart" section of
128 http://archives.seul.org/or/dev/May-2007/msg00008.html
129 o timeout and retry schedules for fetching bridge descriptors
130 - give extend_info_t a router_purpose again
131 o react faster to download networkstatuses after the first bridge
133 o be more robust to bridges being marked as down and leaving us
134 stranded without any known "running" bridges.
135 N . Cache for bridge descriptors
136 o Annotated router store
137 o Accept annotations before routers
138 o Preserve and ignore unexpected annotations
139 o Mechanism to add annotations when we first add a descriptor
140 o Don't serve annotations
141 o Reject annotations that appear in things we've downloaded
142 o Name the router store something different: cached-descriptors?
143 o But load from cached-routers if no cached-descriptors is
146 o Add a few example annotations to make sure this works: source
147 and downloaded/uploaded-at seem like a good start
148 - Drop this later as needed.
149 o Use annotations to denote router purpose
150 o Learn purpose from annotations
151 o Set annotations based on purpose
152 o Preserve routers with unrecognized purpose.
153 R - Stop setting the do-not-cache flag based on purpose.
154 - Disable and mark as obsolete the setrouterpurpose controller
156 - Bridges operators (rudimentary version)
157 - Ability to act as dir cache without a dir port.
158 o Bridges publish to bridge authorities
159 o Fix BEGIN_DIR so that you connect to bridge of which you only
160 know IP (and optionally fingerprint), and then use BEGIN_DIR to learn
162 - look at server_mode() and decide if it always applies to bridges too.
163 - Bridges authorities (rudimentary version)
164 o Rudimentary "do not publish networkstatus" option for bridge
166 - Clients can ask bridge authorities for more bridges.
168 o Clients can ask bridge authorities for updates on known bridges.
169 - More TLS normalization work: make Tor less easily
171 - Directory system improvements
173 - Make BEGIN_DIR mandatory for asking questions of bridge authorities?
175 - Features (other than bridges):
176 - Blocking-resistance.
177 - Write a proposal; make this part of 105.
178 - Audit how much RAM we're using for buffers and cell pools; try to
180 o Accept \n as end of lines in the control protocol in addition to \r\n.
181 o Use fetch_from_buf_line_lf in control.c instead of fetch_from_buf_line.
182 o Fix up read escaped_data to accept LF instead of CRLF, and to
183 always translate_newlines (since that's the only way it's called).
184 - Base relative control socket paths on datadir.
185 - We should ship with a list of stable dir mirrors -- they're not
186 trusted like the authorities, but they'll provide more robustness
187 and diversity for bootstrapping clients.
188 - Better estimates in the directory of whether servers have good uptime
189 (high expected time to failure) or good guard qualities (high
191 - AKA Track uptime as %-of-time-up, as well as time-since-last-down
193 - Make uptime info persist too.
194 - Base Guard on weighted fractional uptime.
195 - Make TrackHostExits expire TrackHostExitsExpire seconds after their
196 *last* use, not their *first* use.
197 - Limit to 2 dir, 2 OR, N SOCKS connections per IP.
198 - Or maybe close connections from same IP when we get a lot from one.
199 - Or maybe block IPs that connect too many times at once.
200 - add an AuthDirBadexit torrc option if we decide we want one.
201 o Add a GuardsSelectedByVersion line to the state file so we know
202 not to drop guards we added.
203 o Have it include the date too.
206 N - Hack up a client that gives out weird/no certificates, so we can
207 test to make sure that this doesn't cause servers to crash.
210 - can we deprecate 'getinfo network-status'?
211 - can we deprecate the FastFirstHopPK config option?
215 - Quietly document NT Service options
216 - More prominently, we should have a recommended apps list.
217 - recommend pidgin (gaim is renamed)
218 - unrecommend IE because of ftp:// bug.
219 - we should add a preamble to tor-design saying it's out of date.
220 . Document transport and natdport in a good HOWTO.
221 - Publicize torel. (What else?
222 . Finish path-spec.txt
225 P - Can we switch to polipo? Please?
226 P - Make documentation realize that location of system configuration file
227 will depend on location of system defaults, and isn't always /etc/torrc.
228 P - If we haven't replaced privoxy, lock down its configuration in all
229 packages, as documented in tor-doc-unix.html
230 P - Figure out why dll's compiled in mingw don't work right in WinXP.
231 P - Create packages for Nokia 800, requested by Chris Soghoian
232 P - Consider creating special Tor-Polipo-Vidalia test packages,
233 requested by Dmitri Vitalev
234 - Get Vidalia supporting protocolinfo and using auth by default.
236 Nice-to-have items for 0.2.0.x, time permitting:
238 - 113: Simplifying directory authority administration
239 - 110: prevent infinite-length circuits (phase one)
240 . Robust decentralized storage for hidden service descriptors.
241 (Karsten is working on this; proposal 114.)
242 - 118: Listen on and advertise multiple ports:
243 - Tor should be able to have a pool of outgoing IP addresses that it is
244 able to rotate through. (maybe. Possible overlap with proposal 118.)
245 - config option to publish what ports you listen on, beyond
246 ORPort/DirPort. It should support ranges and bit prefixes (?) too.
247 (This is very similar to proposal 118.)
249 - Internal code support for ipv6:
250 o Clone ipv6 functions (inet_ntop, inet_pton) where they don't exist.
251 - Most address variables need to become tor_addr_t
252 - Teach resolving code how to handle ipv6.
253 - Teach exit policies about ipv6 (consider ipv4/ipv6 interaction!)
256 - Let controller set router flags for authority to transmit, and for
258 - add an 'exit-address' line in the descriptor for servers that exit
259 from something that isn't their published address.
260 - Clients should estimate their skew as median of skew from servers
262 - More work on AvoidDiskWrites?
265 - MAYBE kill stalled circuits rather than stalled connections. This is
266 possible thanks to cell queues, but we need to consider the anonymity
268 - Implement TLS shutdown properly when possible.
271 - we try to build 4 test circuits to break them over different
272 servers. but sometimes our entry node is the same for multiple
273 test circuits. this defeats the point.
274 - If the client's clock is too far in the past, it will drop (or just not
275 try to get) descriptors, so it'll never build circuits.
278 - Move all status info out of routerinfo into local_routerstatus. Make
279 "who can change what" in local_routerstatus explicit. Make
280 local_routerstatus (or equivalent) subsume all places to go for "what
284 - Detect correct version of libraries from autoconf script.
287 - Review torrc.sample to make it more discursive.
289 Deferred from 0.2.0.x:
293 - Make resolves no longer use edge_connection_t unless they are actually
294 _on_ a socks connection: have edge_connection_t and (say)
295 dns_request_t both extend an edge_stream_t, and have p_streams and
296 n_streams both be linked lists of edge_stream_t.
297 - Generate torrc.{complete|sample}.in, tor.1.in, the HTML manual, and the
298 online config documentation from a single source.
299 - Blocking/scanning-resistance
300 - It would be potentially helpful to https requests on the OR port by
301 acting like an HTTPS server.
302 - Do we want to maintain our own set of entryguards that we use as
303 next hop after the bridge? Open research question; let's say no
304 for 0.2.0 unless we learn otherwise.
305 - Should do reachability testing but only on the purpose==bridge
307 - Some mechanism for specifying that we want to stop using a cached
312 - See also Flyspray tasks.
313 - See also all OPEN/ACCEPTED proposals.
314 - See also all items marked XXXX and FFFF in the code.
317 - Our current approach to block attempts to use Tor as a single-hop proxy
318 is pretty lame; we should get a better one.
319 - Allow small cells and large cells on the same network?
320 - Cell buffering and resending. This will allow us to handle broken
321 circuits as long as the endpoints don't break, plus will allow
322 connection (tls session key) rotation.
323 - Implement Morphmix, so we can compare its behavior, complexity,
324 etc. But see paper breaking morphmix.
325 - Other transport. HTTP, udp, rdp, airhook, etc. May have to do our own
326 link crypto, unless we can bully DTLS into it.
327 - Need a relay teardown cell, separate from one-way ends.
328 (Pending a user who needs this)
329 - Handle half-open connections: right now we don't support all TCP
330 streams, at least according to the protocol. But we handle all that
331 we've seen in the wild.
332 (Pending a user who needs this)
336 - turn the received socks addr:port into a digest for setting .exit
337 - handle connect-dir streams that don't have a chosen_exit_name set.
338 - Have a "Faster" status flag that means it. Fast2, Fast4, Fast8?
339 - Add an option (related to AvoidDiskWrites) to disable directory
340 caching. (Is this actually a good idea??)
341 - Add d64 and fp64 along-side d and fp so people can paste status
342 entries into a url. since + is a valid base64 char, only allow one
343 at a time. Consider adding to controller as well.
344 - Some back-out mechanism for auto-approval on authorities
345 - a way of rolling back approvals to before a timestamp
346 - Consider minion-like fingerprint file/log combination.
347 - Have new people be in limbo and need to demonstrate usefulness
348 before we approve them.
351 - Standby/hotswap/redundant hidden services.
352 . Update the hidden service stuff for the new dir approach. (Much
353 of this will be superseded by 114.)
354 - switch to an ascii format, maybe sexpr?
355 - authdirservers publish blobs of them.
356 - other authdirservers fetch these blobs.
357 - hidserv people have the option of not uploading their blobs.
358 - you can insert a blob via the controller.
359 - and there's some amount of backwards compatibility.
360 - teach clients, intro points, and hidservs about auth mechanisms.
361 - come up with a few more auth mechanisms.
362 - auth mechanisms to let hidden service midpoint and responder filter
364 - Let each hidden service (or other thing) specify its own
366 - Hidserv offerers shouldn't need to define a SocksPort
369 - When we notice a 'Rejected: There is already a named server with
370 this nickname' message... or maybe instead when we see in the
371 networkstatuses that somebody else is Named with the name we
372 want: warn the user, send a STATUS_SERVER message, and fall back
374 - If the server is spewing complaints about raising your ulimit -n,
375 we should add a note about this to the server descriptor so other
376 people can notice too.
377 - When we hit a funny error from a dir request (eg 403 forbidden),
378 but tor is working and happy otherwise, and we haven't seen many
379 such errors recently, then don't warn about it.
382 - A way to adjust router flags from the controller. (How do we
383 prevent the authority from clobbering them soon afterward?)
384 - Implement missing status events and accompanying getinfos
386 - BAD_DIR_RESPONSE (Unexpected directory response; maybe we're behind
388 - BAD_PROXY (Bad http or https proxy)
389 - UNRECOGNIZED_ROUTER (a nickname we asked for is unavailable)
390 - Status events related to hibernation
391 - something about failing to parse our address?
392 from resolve_my_address() in config.c
393 - sketchy OS, sketchy threading
394 - too many onions queued: threading problems or slow CPU?
395 - Implement missing status event fields:
396 - TIMEOUT on CHECKING_REACHABILITY
397 - GETINFO status/client, status/server, status/general: There should be
398 some way to learn which status events are currently "in effect."
399 We should specify which these are, what format they appear in, and so
401 - More information in events:
402 - Include bandwidth breakdown by conn->type in BW events.
403 - Change circuit status events to give more details, like purpose,
404 whether they're internal, when they become dirty, when they become
405 too dirty for further circuits, etc.
406 - Change stream status events analogously.
407 - Expose more information via getinfo:
408 - import and export rendezvous descriptors
409 - Review all static fields for additional candidates
410 - Allow EXTENDCIRCUIT to unknown server.
411 - We need some way to adjust server status, and to tell tor not to
412 download directories/network-status, and a way to force a download.
413 - Make everything work with hidden services
415 - Performance/resources
416 - per-conn write buckets
417 - separate config options for read vs write limiting
418 (It's hard to support read > write, since we need better
419 congestion control to avoid overfull buffers there. So,
420 defer the whole thing.)
421 - Investigate RAM use in directory authorities.
422 - Look into pulling serverdescs off buffers as they arrive.
423 - Rate limit exit connections to a given destination -- this helps
424 us play nice with websites when Tor users want to crawl them; it
425 also introduces DoS opportunities.
426 - Consider truncating rather than destroying failed circuits,
427 in order to save the effort of restarting. There are security
428 issues here that need thinking, though.
429 - Handle full buffers without totally borking
430 - Rate-limit OR and directory connections overall and per-IP and
434 - Hold-open-until-flushed now works by accident; it should work by
436 - Display the reasons in 'destroy' and 'truncated' cells under
438 - Make router_is_general_exit() a bit smarter once we're sure what
440 - Automatically determine what ports are reachable and start using
441 those, if circuits aren't working and it's a pattern we
442 recognize ("port 443 worked once and port 9001 keeps not
446 - don't do dns hijacking tests if we're reject *:* exit policy?
447 (deferred until 0.1.1.x is less common)
450 - Servers might check certs for known-good ssl websites, and if
451 they come back self-signed, declare themselves to be
452 non-exits. Similar to how we test for broken/evil dns now.
453 - Authorities should try using exits for http to connect to some
454 URLS (specified in a configuration file, so as not to make the
455 List Of Things Not To Censor completely obvious) and ask them
456 for results. Exits that don't give good answers should have
457 the BadExit flag set.
458 - Alternatively, authorities should be able to import opinions
459 from Snakes on a Tor.
460 - More consistent error checking in router_parse_entry_from_string().
461 I can say "banana" as my bandwidthcapacity, and it won't even squeak.
462 - Bind to random port when making outgoing connections to Tor servers,
463 to reduce remote sniping attacks.
464 - Audit everything to make sure rend and intro points are just as
465 likely to be us as not.
466 - Do something to prevent spurious EXTEND cells from making
467 middleman nodes connect all over. Rate-limit failed
468 connections, perhaps?
469 - DoS protection: TLS puzzles, public key ops, bandwidth exhaustion.
472 - Tolerate clock skew on bridge relays.
475 - Now that we're avoiding exits when picking non-exit positions,
476 we need to consider how to pick nodes for internal circuits. If
477 we avoid exits for all positions, we skew the load balancing. If
478 we accept exits for all positions, we leak whether it's an
479 internal circuit at every step. If we accept exits only at the
480 last hop, we reintroduce Lasse's attacks from the Oakland paper.
482 - Windows server usability
483 - Solve the ENOBUFS problem.
484 - make tor's use of openssl operate on buffers rather than sockets,
485 so we can make use of libevent's buffer paradigm once it has one.
486 - make tor's use of libevent tolerate either the socket or the
487 buffer paradigm; includes unifying the functions in connect.c.
488 - We need a getrlimit equivalent on Windows so we can reserve some
489 file descriptors for saving files, etc. Otherwise we'll trigger
490 asserts when we're out of file descriptors and crash.
491 - Merge code from Urz into libevent
492 - Make Tor use evbuffers.
495 - a way to generate the website diagrams from source, so we can
496 translate them as utf-8 text rather than with gimp. (svg? or
498 . Flesh out options_description array in src/or/config.c
499 . multiple sample torrc files
500 . figure out how to make nt service stuff work?
502 - Refactor tor man page to divide generally useful options from
504 - Add a doxygen style checker to make check-spaces so nick doesn't drift
505 too far from arma's undocumented styleguide. Also, document that
506 styleguide in HACKING. (See r9634 for example.)
507 - exactly one space at beginning and at end of comments, except i
508 guess when there's line-length pressure.
509 - if we refer to a function name, put a () after it.
510 - only write <b>foo</b> when foo is an argument to this function.
511 - doxygen comments must always end in some form of punctuation.
512 - capitalize the first sentence in the doxygen comment, except
514 - avoid spelling errors and incorrect comments. ;)
517 - The Debian package now uses --verify-config when (re)starting,
518 to distinguish configuration errors from other errors. Perhaps
519 the RPM and other startup scripts should too?
520 - add a "default.action" file to the tor/vidalia bundle so we can
521 fix the https thing in the default configuration:
522 http://wiki.noreply.org/noreply/TheOnionRouter/TorFAQ#PrivoxyWeirdSSLPort
525 - Patch privoxy and socks protocol to pass strings to the browser.
528 Documentation, non-version-specific.
530 - Mark up spec; note unclear points about servers
531 NR - write a spec appendix for 'being nice with tor'
532 - Specify the keys and key rotation schedules and stuff
533 - Mention controller libs someplace.
534 - Remove need for HACKING file.
535 P - document http://wiki.noreply.org/noreply/TheOnionRouter/TransparentProxy on freebsd and osx
536 P - figure out why x86_64 won't build rpms from tor.spec
537 P - figure out spec files for bundles of vidalia-tor-polipo
538 P - figure out polipo install scripts for bundles of vidalia-tor-polipo on osx, win32
539 P - figure out selinux policy for tor
540 P - change packaging system to more automated and specific for each
541 platform, suggested by Paul Wouter
542 P - Setup repos for redhat and suse rpms & start signing the rpms the
543 way package management apps prefer
546 P - tor-in-the-media page
547 - more pictures from ren. he wants to describe the tor handshake
548 - Figure out licenses for website material.
549 P - and remove home and make the "Tor" picture be the link to home.
550 P - put the logo on the website, in source form, so people can put it on
551 stickers directly, etc.
552 P - put the source image for the stickers on the website, so people can
554 P - figure out a license for the logos and docs we publish
555 R - make a page with the hidden service diagrams.
556 P - ask Jan/Jens to be the translation coordinator? add to volunteer page.
557 - add a page for localizing all tor's components.
558 - It would be neat if we had a single place that described _all_ the
559 tor-related tools you can use, and what they give you, and how well they
560 work. Right now, we don't give a lot of guidance wrt
561 torbutton/foxproxy/privoxy/polipo in any consistent place.