[freenet-dev] Current coding priorities

[Matthew Toseland]

Current Priorities:
- Sort out the pre-emptive rejection code. See bug #1245 : The basic
  idea is to not limit the number of requests directly, but limit
  limited resources such as threads and memory, and ensure that even if
  all requests in flight succeed, they will all be able to transfer
  their data within the request timeout period. 
  Work already done: Increasing timeouts, thread limit (thx nextgens)
  and removing old hard limits, refactoring.
- Resource usage and crash-related work: Many people's nodes still crash
  frequently. This is *probably* a memory usage problem, and there is
  more that can be done to reduce Fred's memory footprint. Also nextgens
  has done some work on reducing thread usage, which should eliminate
  some crashes when inserting large freesites, as well as being
  important in the light of the first item.
- The proposal to add UIDs to swap requests in order to investigate
  location clustering, network size, network churn, routing, and
  especially the network topology properly. This adds a minor
  vulnerability by making it easier to map the network, however it is
  only a minor vulnerability as the network close to you can probably be
  mapped fairly easily anyway, and the network close to you is mostly
  the part you want to attack. It would make identifying nodes easier
  with some of the more sophisticated attacks. The purpose of this is to
  make it easy for us to map the network and determine whether the
  current pseudo-opennet is in fact the source of all freenet's ills.
- STS. IMHO shipping Freenet 0.7 as the default without STS makes it
  rather less secure than can reasonably be expected by a user. It is a
  real vulnerability: Man-in-the-Middle attacks or impersonation, if the
  attacker has both noderefs (which he does on pseudo-opennet or real
  opennet), unlike the previous item. And it shouldn't be a large amount
  of code.
- Opennet: Opennet remains essential IMHO to get more users. Right now
  we have lots of potential users who don't get onto Freenet at all. But
  more work should be done on load and routing before it is deployed.
- Darknet: There is some debate going on about renaming the darknet page
  to "friends" or something similar, and making it behave a bit more
  like a buddy list. If somebody would investigate how P2P apps in
  general integrate the buddy list this would be very useful input.
  Nextgens is working on a Jabber plugin to make it easy to invite your
  friends to the darknet (and probably many other useful things e.g. if
  you put a jabber ID in your noderef you can rendezvous when your IP
  address changes; another proposal is IM notification when downloads
  complete).
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://emu.freenetproject.org/pipermail/devl/attachments/20070324/b86ce3f3/attachment.pgp