tor-parallel-relay-conn/doc/TODO

$Id$
Legend:
SPEC!!  - Not specified
SPEC    - Spec not finalized
NICK    - nick claims
ARMA    - arma claims
PHOBOS  - phobos claims
        - Not done
        * Top priority
        . Partially done
        o Done
        D Deferred
        X Abandoned

Non-Coding, Soon:
  - FAQ entry: why gnutls is bad/not good for tor
P - flesh out the rest of the section 6 of the faq
P - gather pointers to livecd distros that include tor
  - put the logo on the website, in source form, so people can put it on
    stickers directly, etc.
  - more pictures from ren. he wants to describe the tor handshake, i want to
    talk about hidden services.
  * clean up the places where our docs are redundant (or worse, obsolete in
    one file and correct elsewhere). agl has a start on a global
    list-of-tor-docs.
P - update window's docs to clarify which versions of windows, and why a
    DOS window, how it's used, for the less technical users
NR- write a spec appendix for 'being nice with tor'
  - Hunt for open socks ports on tor servers, send mail
  - tor-in-the-media page
  - Ask schanzle@cas.homelinux.org about a patch for  rpm spec fixes against
    tor-0.1.0.7.rc
  - Remove need for HACKING file.


For 0.1.0.x:
  o Why do solaris cpuworks go dormant?
    (Apparently, disabling threads fixes this.)
  o Fix the remaining flyspray bugs marked for 0.1.0.9
  X Free remaining unfreed memory (arma will run valgrind)
    (Not for a stable release)
  o Note libevent/method/platform combos that are unlikely to work.
  X change torrc to point to abuse-faq (once abuse-faq is posted)
  . Memory use on Linux: what's happening?
    - Is it threading?  (Maybe, maybe not)
    - Is it the buf_shrink bug? (Quite possibly)
    - Instrument the 0.1.1 code to figure out where our memory is going;
      apply the results. (all platforms?)
  . Why does kevent barf with EINVAL on some freebsd boxes?
    o Submit libevent patch to Niels
    o Warn on non-repeated EINVAL in Tor (don't die.)
    - Investigate why freebsd kernel actually does this: it doesn't seem
      simple to trigger.

for 0.1.1.x:
N . Controller improvements
    o new controller protocol
      o Specify
      o Implement
      . Test, debug
    o add new getinfo options to enumerate things we only find out about
      currently via events.
    o controller should have an event to learn about new addressmappings,
      e.g. when we learn a hostname to IP mapping ?
    . make sure err-level log events get flushed to the controller
      immediately, since tor will exit right after.
      o Implement
      - Test, debug
    - Allow EXTENDCIRCUIT to unknown server.
    . Switch example controllers to use new protocol
      . Python
        o Implement main controller interface
        - Glue code
        - Testing
      - Java
  - switch accountingmax to count total in+out, not either in or
    out. it's easy to move in this direction (not risky), but hard to
    back, out if we decide we prefer it the way it already is. hm.
  . Come up with a coherent strategy for bandwidth buckets and TLS. (The
    logic for reading from TLS sockets is likely to overrun the bandwidth
    buckets under heavy load.  (Really, the logic was never right in the
    first place.)  Also, we should audit all users of get_pending_bytes().)
    - Make it harder to circumvent bandwidth caps: look at number of bytes
      sent across sockets, not number sent inside TLS stream.
N - Handle rendezvousing with unverified nodes.
    o Specify: Stick rendezvous point's address and port in INTRODUCE cell.
    - Handle new format.
      - Parse and generate new format (#if0 out the logic to generate it.)
      - Support to extend circuit/target circuit to a chosen combination of
        addr/port/ID/onionkey
    - Verify that new code works.
  - it looks like tor_assert writes to stderr. what happens if
    stderr was closed and is now something else? uh.
  - christian grothoff's attack of infinite-length circuit.
    the solution is to have a separate 'extend-data' cell type
    which is used for the first N data cells, and only
    extend-data cells can be extend requests.
    - Specify, including thought about
    - Implement
  - Destroy and truncated cells should have reasons.
  - Add private:* alias in exit policies to make it easier to ban all the
    fiddly little 192.168.foo addresses.
    (AGL had a patch; consider applying it.)
  - recommended-versions for client / server ?
  - warn if listening for SOCKS on public IP.
  - Forward-compatibility: add "needclientversion" option or "opt critical"
    prefix.
  - cpu fixes:
    - see if we should make use of truncate to retry
    o hardware accelerator support (configure engines.)
    - hardware accelerator support (use instead of aes.c when reasonable)
r   - kill dns workers more slowly
  - continue decentralizing the directory
    - Specify and design all of the below before implementing any.
    - Figure out what to do about hidden service descriptors.
    M have two router descriptor formats
    - dirservers verify reachability claims
    - find 10 dirservers. (what are criteria to be a dirserver?)
    - some back-out mechanism?
      - dirservers have blacklist of IPs they hate
      - a way of rolling back approvals to before a timestamp
      - have new people be in limbo and need to demonstrate usefulness
        before we approve them
      - other?
    - dirservers publish router-status with all these flags.
    - Servers publish new descriptors when options change, when 12-24 hours
      have passed, when uptime is reset, or when bandwidth changes a lot.
    - alices fetch many router-statuses and update descriptors as needed.
    - add if-newer-than fetch options
    - dirservers allow people to lookup by N descriptors, or to fetch all.
    - alices avoid duplicate class C nodes.
    - everybody with a dirport will give you his descriptor.
    - config option, on by default, to cache all descriptors.
    - Compress router desc sets before transmitting them
    M Analyze how bad the partitioning is or isn't.
  - Naming:
    - Specify and design all of the below before implementing any.
    - some dirservers announce that they manage bindings (a flag in
      router-status).
    - other dirservers mention a binding if there is no conflict for
      that binding among the dirservers that manage it.
      no conflict == any of them bind it and no disagreement.
    - alice can specify a nickname and it will record that name in her
      datadir along with the key *if* it is bound. otherwise her specifying
      will fail (loudly we hope).
    - thus when a binding vanishes (e.g. conflict) alice will keep using
      the one she meant.
    - if the binding changes keys, the entry in her datadir will silently
      get corrected.
  - packaging and ui stuff:
    - multiple sample torrc files (tyranix?)
    - uninstallers
      . for os x
    - something, anything, for sys tray on Windows.
    - figure out how to make nt service stuff work?
      . Document it.
N   - Vet all pending installer patches
      - Win32 installer plus privoxy, sockscap/freecap, etc.
      - Vet win32 systray helper code
N  . Make logs go into platform default locations.
     o OSX
     - Windows. (?)

Reach (deferrable) items for 0.1.1.x:
  - Start using create-fast cells as clients
  - Let more config options (e.g. ORPort) change dynamically.
  - start handling server descriptors without a socksport?

For 0.1.1.x, if we can figure out how:
  - rewrite how libevent does select() on win32 so it's not so very slow.
  - helper nodes (at least preliminary)
  - enclaves (at least preliminary)
  - Write limiting; separate token bucket for write
  - Audit everything to make sure rend and intro points are just as likely to
    be us as not.
  - Do something to prevent spurious EXTEND cells from making middleman
    nodes connect all over.  Rate-limit failed connections, perhaps?

Future version:
  - Limit to 2 dir, 2 OR, N SOCKS connections per IP.
  - Handle full buffers without totally borking
  - Rate-limit OR and directory connections overall and per-IP and
    maybe per subnet.
  - Hold-open-until-flushed now works by accident; it should work by
    design.
  - DoS protection: TLS puzzles, public key ops, bandwidth exhaustion.
   - Specify?
  - tor-resolve script should use socks5 to get better error messages.
  - make min uptime a function of the available choices (say, choose 60th
    percentile, not 1 day.)
  - config option to publish what ports you listen on, beyond ORPort/DirPort
  - hidserv offerers shouldn't need to define a SocksPort
    * figure out what breaks for this, and do it.
  - auth mechanisms to let hidden service midpoint and responder filter
    connection requests.
  - Relax clique assumptions.
  - tor should be able to have a pool of outgoing IP addresses
    that it is able to rotate through. (maybe)

Blue-sky:
  - Patch privoxy and socks protocol to pass strings to the browser.
  - Standby/hotswap/redundant hidden services.
  - Robust decentralized storage for hidden service descriptors.
  - The "China problem"
  - Allow small cells and large cells on the same network?
  - Cell buffering and resending. This will allow us to handle broken
    circuits as long as the endpoints don't break, plus will allow
    connection (tls session key) rotation.
  - Implement Morphmix, so we can compare its behavior, complexity, etc.
  - Other transport. HTTP, udp, rdp, airhook, etc. May have to do our own
    link crypto, unless we can bully openssl into it.
  - Conn key rotation.
  - Need a relay teardown cell, separate from one-way ends.
    (Pending a user who needs this)
  - Handle half-open connections: right now we don't support all TCP
    streams, at least according to the protocol. But we handle all that
    we've seen in the wild.
    (Pending a user who needs this)