4 SPEC - Spec not finalized
16 N - contact umass folks
17 N - Packaging logic and HOWTO for controller libs
18 N - Mention controller libs someplace.
19 - FAQ entry: why gnutls is bad/not good for tor
20 P - flesh out the rest of the section 6 of the faq
21 P - gather pointers to livecd distros that include tor
22 - put the logo on the website, in source form, so people can put it on
23 stickers directly, etc.
24 - more pictures from ren. he wants to describe the tor handshake, i want to
25 talk about hidden services.
26 * clean up the places where our docs are redundant (or worse, obsolete in
27 one file and correct elsewhere). agl has a start on a global
29 P - update window's docs to clarify which versions of windows, and why a
30 DOS window, how it's used, for the less technical users
31 NR- write a spec appendix for 'being nice with tor'
32 - Hunt for open socks ports on tor servers, send mail
33 - tor-in-the-media page
34 - Ask schanzle@cas.homelinux.org about a patch for rpm spec fixes against
36 - Remove need for HACKING file.
40 o Why do solaris cpuworks go dormant?
41 (Apparently, disabling threads fixes this.)
42 o Fix the remaining flyspray bugs marked for 0.1.0.9
43 X Free remaining unfreed memory (arma will run valgrind)
44 (Not for a stable release)
45 o Note libevent/method/platform combos that are unlikely to work.
46 X change torrc to point to abuse-faq (once abuse-faq is posted)
47 . Memory use on Linux: what's happening?
48 - Is it threading? (Maybe, maybe not)
49 - Is it the buf_shrink bug? (Quite possibly)
50 - Instrument the 0.1.1 code to figure out where our memory is going;
51 apply the results. (all platforms?)
52 . Why does kevent barf with EINVAL on some freebsd boxes?
53 o Submit libevent patch to Niels
54 o Warn on non-repeated EINVAL in Tor (don't die.)
55 - Investigate why freebsd kernel actually does this: it doesn't seem
59 N . Controller improvements
60 o new controller protocol
64 o add new getinfo options to enumerate things we only find out about
66 o controller should have an event to learn about new addressmappings,
67 e.g. when we learn a hostname to IP mapping ?
68 . make sure err-level log events get flushed to the controller
69 immediately, since tor will exit right after.
72 - Allow EXTENDCIRCUIT to unknown server.
73 . Switch example controllers to use new protocol
75 o Implement main controller interface
79 o Implement main controller interface
82 - switch accountingmax to count total in+out, not either in or
83 out. it's easy to move in this direction (not risky), but hard to
84 back, out if we decide we prefer it the way it already is. hm.
85 . Come up with a coherent strategy for bandwidth buckets and TLS. (The
86 logic for reading from TLS sockets is likely to overrun the bandwidth
87 buckets under heavy load. (Really, the logic was never right in the
88 first place.) Also, we should audit all users of get_pending_bytes().)
89 - Make it harder to circumvent bandwidth caps: look at number of bytes
90 sent across sockets, not number sent inside TLS stream.
91 N . Handle rendezvousing with unverified nodes.
92 o Specify: Stick rendezvous point's address and port in INTRODUCE cell.
94 o Support to extend circuit/target circuit to a chosen combination of
97 o Generate new format (#ifdef out the logic to generate it for now)
98 o Specify: make service descriptors contain onion key and identity.
99 o Implement new service desc format
100 - Think: are we okay with the partitioning?
101 o Implement new directory code
102 o Implement new server code (Don't enable till directory code is deployed)
103 o Implement new client code (Don't enable till directory code is deployed)
104 o Look for v1 descriptor if available, else look for v0 descriptor.
105 o Use new INTRODUCE protocol if allowed.
106 . Verify that new code works.
107 X It looks like tor_assert writes to stderr. This isn't a problem, because
108 start_daemon doesn't close fd 2; it uses dup2 to replace it with
109 a file open to /dev/null.
110 - christian grothoff's attack of infinite-length circuit.
111 the solution is to have a separate 'extend-data' cell type
112 which is used for the first N data cells, and only
113 extend-data cells can be extend requests.
114 - Specify, including thought about
116 - Destroy and truncated cells should have reasons.
117 - Add private:* alias in exit policies to make it easier to ban all the
118 fiddly little 192.168.foo addresses.
119 (AGL had a patch; consider applying it.)
120 - recommended-versions for client / server ?
121 - warn if listening for SOCKS on public IP.
122 o Forward-compatibility: add "needclientversion" option or "opt critical"
123 prefix? No, just make unknown keywords less critical.
125 - see if we should make use of truncate to retry
126 o hardware accelerator support (configure engines.)
127 - hardware accelerator support (use instead of aes.c when reasonable)
128 r - kill dns workers more slowly
129 - continue decentralizing the directory
130 - Specify and design all of the below before implementing any.
131 - Figure out what to do about hidden service descriptors.
132 M have two router descriptor formats
133 - dirservers verify reachability claims
134 - find 10 dirservers. (what are criteria to be a dirserver?)
135 - some back-out mechanism?
136 - dirservers have blacklist of IPs they hate
137 - a way of rolling back approvals to before a timestamp
138 - have new people be in limbo and need to demonstrate usefulness
139 before we approve them
141 - dirservers publish router-status with all these flags.
142 - Servers publish new descriptors when options change, when 12-24 hours
143 have passed, when uptime is reset, or when bandwidth changes a lot.
144 - alices fetch many router-statuses and update descriptors as needed.
145 - add if-newer-than fetch options
146 - dirservers allow people to lookup by N descriptors, or to fetch all.
147 - alices avoid duplicate class C nodes.
148 - everybody with a dirport will give you his descriptor.
149 - config option, on by default, to cache all descriptors.
150 - Compress router desc sets before transmitting them
151 M Analyze how bad the partitioning is or isn't.
153 - Specify and design all of the below before implementing any.
154 - some dirservers announce that they manage bindings (a flag in
156 - other dirservers mention a binding if there is no conflict for
157 that binding among the dirservers that manage it.
158 no conflict == any of them bind it and no disagreement.
159 - alice can specify a nickname and it will record that name in her
160 datadir along with the key *if* it is bound. otherwise her specifying
161 will fail (loudly we hope).
162 - thus when a binding vanishes (e.g. conflict) alice will keep using
164 - if the binding changes keys, the entry in her datadir will silently
166 - packaging and ui stuff:
167 - multiple sample torrc files (tyranix?)
170 - something, anything, for sys tray on Windows.
171 - figure out how to make nt service stuff work?
173 - Simple logic to estimate number of active/total users
174 - Add version number to directory.
175 N - Vet all pending installer patches
176 - Win32 installer plus privoxy, sockscap/freecap, etc.
177 - Vet win32 systray helper code
178 N . Make logs go into platform default locations.
182 Reach (deferrable) items for 0.1.1.x:
183 - Start using create-fast cells as clients
184 - Let more config options (e.g. ORPort) change dynamically.
185 - start handling server descriptors without a socksport?
187 For 0.1.1.x, if we can figure out how:
188 - rewrite how libevent does select() on win32 so it's not so very slow.
189 - helper nodes (at least preliminary)
190 - enclaves (at least preliminary)
191 - Write limiting; separate token bucket for write
192 - Audit everything to make sure rend and intro points are just as likely to
194 - Do something to prevent spurious EXTEND cells from making middleman
195 nodes connect all over. Rate-limit failed connections, perhaps?
198 - Limit to 2 dir, 2 OR, N SOCKS connections per IP.
199 - Handle full buffers without totally borking
200 - Rate-limit OR and directory connections overall and per-IP and
202 - Hold-open-until-flushed now works by accident; it should work by
204 - DoS protection: TLS puzzles, public key ops, bandwidth exhaustion.
206 - tor-resolve script should use socks5 to get better error messages.
207 - make min uptime a function of the available choices (say, choose 60th
208 percentile, not 1 day.)
209 - config option to publish what ports you listen on, beyond ORPort/DirPort
210 - hidserv offerers shouldn't need to define a SocksPort
211 * figure out what breaks for this, and do it.
212 - auth mechanisms to let hidden service midpoint and responder filter
214 - Relax clique assumptions.
215 - tor should be able to have a pool of outgoing IP addresses
216 that it is able to rotate through. (maybe)
219 - Patch privoxy and socks protocol to pass strings to the browser.
220 - Standby/hotswap/redundant hidden services.
221 - Robust decentralized storage for hidden service descriptors.
222 - The "China problem"
223 - Allow small cells and large cells on the same network?
224 - Cell buffering and resending. This will allow us to handle broken
225 circuits as long as the endpoints don't break, plus will allow
226 connection (tls session key) rotation.
227 - Implement Morphmix, so we can compare its behavior, complexity, etc.
228 - Other transport. HTTP, udp, rdp, airhook, etc. May have to do our own
229 link crypto, unless we can bully openssl into it.
231 - Need a relay teardown cell, separate from one-way ends.
232 (Pending a user who needs this)
233 - Handle half-open connections: right now we don't support all TCP
234 streams, at least according to the protocol. But we handle all that
235 we've seen in the wild.
236 (Pending a user who needs this)