TODO 25 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537
  1. $Id$
  2. Legend:
  3. SPEC!! - Not specified
  4. SPEC - Spec not finalized
  5. N - nick claims
  6. R - arma claims
  7. P - phobos claims
  8. - Not done
  9. * Top priority
  10. . Partially done
  11. o Done
  12. d Deferrable
  13. D Deferred
  14. X Abandoned
  15. X <nickm> "Let's try to find a way to make it run and make the version
  16. match, but if not, let's just make it run."
  17. X <arma> "should we detect if we have a --with-ssl-dir and try the -R
  18. by default, if it works?"
  19. Items for 0.1.2.x, real soon now:
  20. ? - Bug: combination of things:
  21. When we've been idle a long time, we stop fetching server
  22. descriptors. When we then get a socks request, we build circuits
  23. immediately using whatever descriptors we have, rather than waiting
  24. until we've fetched correct ones.
  25. N - Test guard unreachable logic; make sure that we actually attempt to
  26. connect to guards that we think are unreachable from time to time.
  27. Make sure that we don't freak out when the network is down.
  28. R - Reconstruct ChangeLog; put rolled-up info in ReleaseNotes or something.
  29. Items for 0.1.2.x:
  30. - weight dir requests by advertised bandwidth? with maybe a lower cutoff
  31. than for tor traffic. perhaps also weighted by the expected size of
  32. the response.
  33. o enumerate events of important things that occur in tor, so vidalia can
  34. react.
  35. o Backend implementation
  36. o Actually list all the events (notice and warn log messages are a good
  37. place to look.) Divide messages into categories, perhaps.
  38. o Specify general event system
  39. o Specify actual events.
  40. o Implement or defer remaining events
  41. D Implement or defer GETINFO list of current status events.
  42. o Clean up relevant bits of control-spec.txt
  43. . Have (and document) a BEGIN_DIR relay cell that means "Connect to your
  44. directory port."
  45. o Implement
  46. R - turn the received socks addr:port into a digest for setting .exit
  47. - be able to connect without having a server descriptor, to bootstrap.
  48. R - handle connect-dir streams that don't have a chosen_exit_name set.
  49. o include ORPort in DirServers lines so we can know where to connect.
  50. list the orport as 0 if it can't handle begin_dir.
  51. - List orports of actual dirservers..
  52. - Servers are easy to setup and run: being a relay is about as easy as
  53. being a client.
  54. . Reduce resource load
  55. o A way to alert controller when router flags change.
  56. o Specify: SETEVENTS NS
  57. o Implement
  58. R - Hunt for places that change networkstatus info that I might have
  59. missed.
  60. . option to dl directory info via tor:
  61. TunnelDirConns and PreferTunneledDirConns
  62. R - actually cause the directory.c functions to know about or_port
  63. and use it when we're supposed to.
  64. N - for tunneled edge conns, stop reading to the bridge connection
  65. when the or_conn we're writing to has a full outbuf.
  66. - This is iffy. Really, it would've been better to stop writing
  67. on the dir conn when the edge conn's inbuf is getting full.
  68. But the issue there is that we package from edge
  69. connections aggressively until we hit their package windows or the
  70. circuit package windows, even if the buffer on the corresponding OR
  71. connection is pretty damn big. This sucks from a RAM usage POV.
  72. Now, we could try to stop reading on the edges (or just the edges
  73. connected to a local bridge) when an or_conn's outbuf is full. But
  74. if we're a server, and we stop reading on some exit conns when OR
  75. conns are full, soon OR conns will contain only traffic from other OR
  76. conns, and the exit data in question will never get written.
  77. - The right solution in the long run (0.2.0) is probably as follows:
  78. - Remove socketpair-based bridges: use shared (or connected) buffers
  79. for communication, rather than sockets.
  80. - When relaying cells from an OR conn to an OR conn, have them wait
  81. in a queue on the or_circuit_t object; don't move them onto the
  82. target conn until there is space in the target conn's outbuf.
  83. Also, only package data from exitconns when there is space in the
  84. target conn's outbuf.
  85. - As an added advantage, this would let us kill stalled _circuits_
  86. when their buffers get too full, rather than killing entire OR
  87. conns. But we must think about anonymity implications of that.
  88. - We'll probably want to do some kind of big refactoring of our
  89. dataflow when we do these changes; stuff is hairy enough already,
  90. and it will only get harrier with this stuff.
  91. - For the short run, our options seem to be:
  92. - Disable BEGIN_DIR support at the server-side unless it's
  93. explicitly turned on.
  94. - Go ahead and make directory bridge data not get produced when the
  95. corresponding or_conn is full, and accept the sometimes directory
  96. data will just never get written.
  97. - Rate-limit directory bridge data somehow when the corresponding
  98. or_conn is full. Possibly based on the flush rate of the or_conn?
  99. - Reject BEGIN_DIR requests that we think will be insanely big.
  100. N - DNS improvements
  101. . Asynchronous DNS
  102. - Make evdns use windows strerror equivalents.
  103. - Make sure patches get into libevent.
  104. - Verify that it works well on windows
  105. - Debug and re-enable server-side reverse DNS caching
  106. - Critical but minor bugs, backport candidates.
  107. - support dir 503s better
  108. o clients don't log as loudly when they receive them
  109. o they don't count toward the 3-strikes rule
  110. D But eventually, we give up after getting a lot of 503s.
  111. D Delay when we get a lot of 503s, rather than punting onto the
  112. servers that have given us 503s?
  113. o Add a 'BadDirectory' flag to statuses.
  114. o authorities should *never* 503 a cache, and should never 503
  115. network status requests.
  116. D They can 503 client descriptor requests when they feel like it.
  117. How can they distinguish? Not implemented for now, maybe
  118. should abandon.
  119. - update dir-spec with what we decided for each of these
  120. o Have a mode that doesn't write to disk much, so we can run Tor on
  121. flash memory (e.g. Linksys routers or USB keys).
  122. o Add AvoidDiskWrites config option.
  123. o only write state file when it's "changed"
  124. o crank up the numbers if avoiddiskwrites is on.
  125. D some things may not want to get written at all.
  126. o stop writing fingerprint every restart
  127. D more?
  128. NR. Write path-spec.txt
  129. - Polishing
  130. - Profile client and server; fix slow spots
  131. - Address XXX012 items
  132. - Packaging
  133. - Tell people about OSX Uninstaller
  134. - Quietly document NT Service options
  135. - Switch canonical win32 compiler to mingw.
  136. NR D Get some kind of "meta signing key" to be used solely to sign
  137. releases/to certify releases when signed by the right people/
  138. to certify sign the right people's keys? Also use this to cert the SSL
  139. key, etc.
  140. - If we haven't replaced privoxy, lock down its configuration in all
  141. packages, as documented in tor-doc-unix.html
  142. - Docs
  143. - More prominently, we should have a recommended apps list.
  144. - recommend gaim.
  145. - unrecommend IE because of ftp:// bug.
  146. N - torrc.complete.in needs attention?
  147. N - we should add a preamble to tor-design saying it's out of date.
  148. N - Document transport and natdport
  149. - Improvements to bandwidth counting
  150. R - look into "uncounting" bytes spent on local connections, so
  151. we can bandwidthrate but still have fast downloads.
  152. R - "bandwidth classes", for incoming vs initiated-here conns,
  153. and to give dir conns lower priority.
  154. . Write limiting; separate token bucket for write
  155. o preemptively give a 503 to some v1 dir requests
  156. o preemptively give a 503 to some v2 dir requests
  157. o Write function to estimate bytes needed for N descriptors
  158. statuses
  159. D per-conn write buckets
  160. D separate config options for read vs write limiting
  161. (It's hard to support read > write, since we need better
  162. congestion control to avoid overfull buffers there. So,
  163. defer the whole thing.)
  164. - Forward compatibility fixes
  165. - Caches should start trying to cache consensus docs?
  166. NR - Design
  167. N - Implement, if we think it's smart.
  168. - Start uploading short and long descriptors; authorities should support
  169. URLs to retrieve long descriptors, and should discard short descriptors
  170. for now. Later, once tools use the "long descriptor" URLs, authorities
  171. will serve the short descriptors every time they're asked for
  172. a descriptor.
  173. NR - Design
  174. N - Implement, if we think it's smart.
  175. Topics to think about during 0.1.2.x development:
  176. * Figure out incentives.
  177. - (How can we make this tolerant of a bad v0?)
  178. * Figure out non-clique.
  179. * Figure out China.
  180. - Figure out partial network knowledge.
  181. - Figure out hidden services.
  182. - Design next-version protocol for directories
  183. - Design next-version protocol for connections
  184. Deferred from 0.1.2.x:
  185. P - Figure out why dll's compiled in mingw don't work right in WinXP.
  186. P - Figure out why openssl 0.9.8d "make test" fails at sha256t test.
  187. - Directory guards
  188. - RAM use in directory authorities.
  189. - Memory use improvements:
  190. - Look into pulling serverdescs off buffers as they arrive.
  191. - Save and mmap v1 directories, and networkstatus docs; store them
  192. zipped, not uncompressed.
  193. - Switch cached_router_t to use mmap.
  194. - What to do about reference counts on windows? (On Unix, this is
  195. easy: unlink works fine. (Right?) On Windows, I have doubts. Do we
  196. need to keep multiple files?)
  197. - What do we do about the fact that people can't read zlib-
  198. compressed files manually?
  199. - If the client's clock is too far in the past, it will drop (or
  200. just not try to get) descriptors, so it'll never build circuits.
  201. - Tolerate clock skew on bridge relays.
  202. - A more efficient dir protocol.
  203. - Authorities should fetch the network-statuses amongst each
  204. other, consensus them, and advertise a communal network-status.
  205. This is not so much for safety/complexity as it is to reduce
  206. bandwidth requirements for Alice.
  207. - How does this interact with our goal of being able to choose
  208. your own dir authorities? I guess we're now assuming that all
  209. dir authorities know all the other authorities in their "group"?
  210. - Should we also look into a "delta since last network-status
  211. checkpoint" scheme, to reduce overhead further?
  212. - Extend the "r" line in network-status to give a set of buckets (say,
  213. comma-separated) for that router.
  214. - Buckets are deterministic based on IP address.
  215. - Then clients can choose a bucket (or set of buckets) to
  216. download and use.
  217. - Improvements to versioning.
  218. - When we connect to a Tor server, it sends back a cell listing
  219. the IP it believes it is using. Use this to block dvorak's attack.
  220. Also, this is a fine time to say what time you think it is.
  221. o Verify that a new cell type is okay with deployed codebase
  222. . Specify HELLO cells
  223. . Figure out v0 compatibility.
  224. - Implement
  225. - Eventdns improvements
  226. - Have a way to query for AAAA and A records simultaneously.
  227. - Improve request API: At the very least, add the ability to construct
  228. a more-or-less arbitrary request and get a response.
  229. - (Can we suppress cnames? Should we?)
  230. - Now that we're avoiding exits when picking non-exit positions,
  231. we need to consider how to pick nodes for internal circuits. If
  232. we avoid exits for all positions, we skew the load balancing. If
  233. we accept exits for all positions, we leak whether it's an internal
  234. circuit at every step. If we accept exits only at the last hop, we
  235. reintroduce Lasse's attacks from the Oakland paper.
  236. - We should ship with a list of stable dir mirrors -- they're not
  237. trusted like the authorities, but they'll provide more robustness
  238. and diversity for bootstrapping clients.
  239. - Simplify authority operation
  240. - Follow weasel's proposal, crossed with mixminion dir config format
  241. - A way to adjust router flags from the controller.
  242. (How do we prevent the authority from clobbering them soon after?)
  243. - a way to pick entry guards based wholly on extend_info equivalent;
  244. a way to export extend_info equivalent.
  245. - Better estimates in the directory of whether servers have good uptime
  246. (high expected time to failure) or good guard qualities (high
  247. fractional uptime).
  248. - AKA Track uptime as %-of-time-up, as well as time-since-last-down
  249. - Have a "Faster" status flag that means it. Fast2, Fast4, Fast8?
  250. - spec
  251. - implement
  252. - Failed rend desc fetches sometimes don't get retried. True/false?
  253. - Windows server usability
  254. - Solve the ENOBUFS problem.
  255. - make tor's use of openssl operate on buffers rather than sockets,
  256. so we can make use of libevent's buffer paradigm once it has one.
  257. - make tor's use of libevent tolerate either the socket or the
  258. buffer paradigm; includes unifying the functions in connect.c.
  259. - We need a getrlimit equivalent on Windows so we can reserve some
  260. file descriptors for saving files, etc. Otherwise we'll trigger
  261. asserts when we're out of file descriptors and crash.
  262. M - rewrite how libevent does select() on win32 so it's not so very slow.
  263. - Add overlapped IO
  264. - Add an option (related to AvoidDiskWrites) to disable directory caching.
  265. - More status event features:
  266. - Missing events:
  267. - DIR_REACHABLE
  268. - BAD_DIR_RESPONSE (Unexpected directory response; maybe we're behind
  269. a firewall.)
  270. - BAD_PROXY (Bad http or https proxy)
  271. - UNRECOGNIZED_ROUTER (a nickname we asked for is unavailable)
  272. - Status events related to hibernation
  273. - something about failing to parse our address?
  274. from resolve_my_address() in config.c
  275. - sketchy OS, sketchy threading
  276. - too many onions queued: threading problems or slow CPU?
  277. - Missing fields:
  278. - TIMEOUT on CHECKING_REACHABILITY
  279. - GETINFO status/client, status/server, status/general: There should be
  280. some way to learn which status events are currently "in effect."
  281. We should specify which these are, what format they appear in, and so
  282. on.
  283. Minor items for 0.1.2.x as time permits:
  284. - include bandwidth breakdown by conn->type in BW events.
  285. - getinfo ns/name/moria2 doesn't include a "v" line, even when some
  286. network-statuses I have show it. I suppose the fix should go in
  287. networkstatus_getinfo_helper_single() in routerlist.c.
  288. - Unify autoconf search code for libevent and openssl. Make code
  289. suggest platform-appropriate "devel" / "dev" / whatever packages
  290. if we can link but we can't find the headers.
  291. - Recommend polipo? Please?
  292. - Make documentation realize that location of system configuration file
  293. will depend on location of system defaults, and isn't always /etc/torrc.
  294. - Review torrc.sample to make it more discursive.
  295. o when reporting clock skew (both to logs and to controller), if it's
  296. taken 126 seconds to read from the directory, our clock skew estimate
  297. is 126 seconds wrong. use conn->timestamp_create or _lastwritten
  298. for a closer estimate?
  299. - a way to generate the website diagrams from source, so we can
  300. translate them as utf-8 text rather than with gimp.
  301. R - add d64 and fp64 along-side d and fp so people can paste status
  302. entries into a url. since + is a valid base64 char, only allow one
  303. at a time. spec and then do.
  304. D don't do dns hijacking tests if we're reject *:* exit policy?
  305. (deferred until 0.1.1.x is less common)
  306. - When we export something from foo.c file for testing purposes only,
  307. make a foo_test.h file for test.c to include.
  308. - The Debian package now uses --verify-config when (re)starting,
  309. to distinguish configuration errors from other errors. Perhaps
  310. the RPM and other startup scripts should too?
  311. - add a "default.action" file to the tor/vidalia bundle so we can fix the
  312. https thing in the default configuration:
  313. http://wiki.noreply.org/noreply/TheOnionRouter/TorFAQ#PrivoxyWeirdSSLPort
  314. . Flesh out options_description array in src/or/config.c
  315. - Don't let 'newnym' be triggered more often than every n seconds.
  316. X If we try to publish as a nickname that's already claimed, should
  317. we append a number (or increment the number) and try again? This
  318. way people who read their logs can fix it as before, but people
  319. who don't read their logs will still offer Tor servers.
  320. - Fall back to unnamed; warn user; send controller event. ("When we
  321. notice a 'Rejected: There is already a named server with this nickname'
  322. message... or maybe instead when we see in the networkstatuses that
  323. somebody else is Named with the name we want: warn the user, send a
  324. STATUS_SERVER message, and fall back to unnamed.")
  325. ! - Tor should bind its ports before dropping privs, so users don't
  326. have to do the ipchains dance.
  327. - Rate limit exit connections to a given destination -- this helps
  328. us play nice with websites when Tor users want to crawl them; it
  329. also introduces DoS opportunities.
  330. - Streamline how we pick entry nodes: Make choose_random_entry() have
  331. less magic and less control logic.
  332. - Christian Grothoff's attack of infinite-length circuit.
  333. the solution is to have a separate 'extend-data' cell type
  334. which is used for the first N data cells, and only
  335. extend-data cells can be extend requests.
  336. - Specify, including thought about anonymity implications.
  337. - Display the reasons in 'destroy' and 'truncated' cells under some
  338. circumstances?
  339. - We need a way for the authorities to declare that nodes are
  340. in a family. Also, it kinda sucks that family declarations use O(N^2)
  341. space in the descriptors.
  342. - If the server is spewing complaints about raising your ulimit -n,
  343. we should add a note about this to the server descriptor so other
  344. people can notice too.
  345. - cpu fixes:
  346. - see if we should make use of truncate to retry
  347. X kill dns workers more slowly
  348. . Directory changes
  349. . Some back-out mechanism for auto-approval
  350. - a way of rolling back approvals to before a timestamp
  351. - Consider minion-like fingerprint file/log combination.
  352. - packaging and ui stuff:
  353. . multiple sample torrc files
  354. . figure out how to make nt service stuff work?
  355. . Document it.
  356. - Vet all pending installer patches
  357. - Win32 installer plus privoxy, sockscap/freecap, etc.
  358. - Vet win32 systray helper code
  359. - Improve controller
  360. - a NEWSTATUS event similar to NEWDESC.
  361. - change circuit status events to give more details, like purpose,
  362. whether they're internal, when they become dirty, when they become
  363. too dirty for further circuits, etc.
  364. - What do we want here, exactly?
  365. - Specify and implement it.
  366. - Change stream status events analogously.
  367. - What do we want here, exactly?
  368. - Specify and implement it.
  369. - Make other events "better".
  370. - Change stream status events analogously.
  371. - What do we want here, exactly?
  372. - Specify and implement it.
  373. - Make other events "better" analogously
  374. - What do we want here, exactly?
  375. - Specify and implement it.
  376. . Expose more information via getinfo:
  377. - import and export rendezvous descriptors
  378. - Review all static fields for additional candidates
  379. - Allow EXTENDCIRCUIT to unknown server.
  380. - We need some way to adjust server status, and to tell tor not to
  381. download directories/network-status, and a way to force a download.
  382. - Make everything work with hidden services
  383. - Directory system improvements
  384. - config option to publish what ports you listen on, beyond
  385. ORPort/DirPort. It should support ranges and bit prefixes (?) too.
  386. - Parse this.
  387. - Relay this in networkstatus.
  388. - Be a DNS proxy.
  389. - Need a way to request address lookups (and allocate a stream ID for
  390. them) without having a corresponding client socket.
  391. - Once this is done, it would be nice to have a way to request address
  392. lookups from the controller without using SOCKS.
  393. - Refactor exit/entry side of DNS resolve: we don't need a connection_t;
  394. we can have an edge_connection_t and (say) dns_request_t both extend an
  395. edge_stream_t, and have p_streams and n_streams both be linked lists
  396. of edge_stream_t.
  397. - Look into generating torrc.{complete|sample}.in, tor.1.in,
  398. the HTML manual, and the online config documentation from a single
  399. source.
  400. Future version:
  401. - Configuration format really wants sections.
  402. - Good RBL substitute.
  403. - Authorities should try using exits for http to connect to some URLS
  404. (specified in a configuration file, so as not to make the List Of Things
  405. Not To Censor completely obvious) and ask them for results. Exits that
  406. don't give good answers should have the BadExit flag set.
  407. - Our current approach to block attempts to use Tor as a single-hop proxy
  408. is pretty lame; we should get a better one.
  409. . Update the hidden service stuff for the new dir approach.
  410. - switch to an ascii format, maybe sexpr?
  411. - authdirservers publish blobs of them.
  412. - other authdirservers fetch these blobs.
  413. - hidserv people have the option of not uploading their blobs.
  414. - you can insert a blob via the controller.
  415. - and there's some amount of backwards compatibility.
  416. - teach clients, intro points, and hidservs about auth mechanisms.
  417. - come up with a few more auth mechanisms.
  418. - auth mechanisms to let hidden service midpoint and responder filter
  419. connection requests.
  420. - Bind to random port when making outgoing connections to Tor servers,
  421. to reduce remote sniping attacks.
  422. - Have new people be in limbo and need to demonstrate usefulness
  423. before we approve them.
  424. - Clients should estimate their skew as median of skew from servers
  425. over last N seconds.
  426. - Make router_is_general_exit() a bit smarter once we're sure what it's for.
  427. - Audit everything to make sure rend and intro points are just as likely to
  428. be us as not.
  429. - Do something to prevent spurious EXTEND cells from making middleman
  430. nodes connect all over. Rate-limit failed connections, perhaps?
  431. - Automatically determine what ports are reachable and start using
  432. those, if circuits aren't working and it's a pattern we recognize
  433. ("port 443 worked once and port 9001 keeps not working").
  434. - Limit to 2 dir, 2 OR, N SOCKS connections per IP.
  435. - Handle full buffers without totally borking
  436. - Rate-limit OR and directory connections overall and per-IP and
  437. maybe per subnet.
  438. - Hold-open-until-flushed now works by accident; it should work by
  439. design.
  440. - DoS protection: TLS puzzles, public key ops, bandwidth exhaustion.
  441. - Specify?
  442. - tor-resolve script should use socks5 to get better error messages.
  443. - hidserv offerers shouldn't need to define a SocksPort
  444. * figure out what breaks for this, and do it.
  445. - tor should be able to have a pool of outgoing IP addresses
  446. that it is able to rotate through. (maybe)
  447. - Specify; implement.
  448. - let each hidden service (or other thing) specify its own
  449. OutboundBindAddress?
  450. - Stop using tor_socketpair to make connection bridges: do an
  451. implementation that uses buffers only.
  452. Blue-sky:
  453. - Patch privoxy and socks protocol to pass strings to the browser.
  454. - Standby/hotswap/redundant hidden services.
  455. - Robust decentralized storage for hidden service descriptors.
  456. - The "China problem"
  457. - Allow small cells and large cells on the same network?
  458. - Cell buffering and resending. This will allow us to handle broken
  459. circuits as long as the endpoints don't break, plus will allow
  460. connection (tls session key) rotation.
  461. - Implement Morphmix, so we can compare its behavior, complexity, etc.
  462. - Other transport. HTTP, udp, rdp, airhook, etc. May have to do our own
  463. link crypto, unless we can bully openssl into it.
  464. - Need a relay teardown cell, separate from one-way ends.
  465. (Pending a user who needs this)
  466. - Handle half-open connections: right now we don't support all TCP
  467. streams, at least according to the protocol. But we handle all that
  468. we've seen in the wild.
  469. (Pending a user who needs this)
  470. Non-Coding:
  471. - Mark up spec; note unclear points about servers
  472. - Mention controller libs someplace.
  473. . more pictures from ren. he wants to describe the tor handshake
  474. NR- write a spec appendix for 'being nice with tor'
  475. - tor-in-the-media page
  476. - Remove need for HACKING file.
  477. - Figure out licenses for website material.
  478. - Specify the keys and key rotation schedules and stuff
  479. Website:
  480. - and remove home and make the "Tor" picture be the link to home.
  481. - put the logo on the website, in source form, so people can put it on
  482. stickers directly, etc.
  483. R - make a page with the hidden service diagrams.
  484. - ask Jan to be the translation coordinator? add to volunteer page.
  485. - add a page for localizing all tor's components.
  486. - It would be neat if we had a single place that described _all_ the
  487. tor-related tools you can use, and what they give you, and how well they
  488. work. Right now, we don't give a lot of guidance wrt
  489. torbutton/foxproxy/privoxy/polipo in any consistent place.