Author Topic: GNUnet 0.7.0pre2 released  (Read 3371 times)

0 Members and 1 Guest are viewing this topic.

Offline crypton

  • Elite
  • *****
  • Posts: 1699
  • Karma: +10/-0
    • View Profile
GNUnet 0.7.0pre2 released
« on: April 04, 2005, 06:54:04 AM »
GNUnet 0.7.0pre2 released

This release fixes a couple of critical bugs discovered in 0.7.0pre1:*

    * potential segfault in the Win32 version*
    * segfault in advertising (crashing nodes under certain circumstances)*
    * gnunet-insert did not work properly when used with multiple keywords (would concatenate all of them into one keyword)*
    * TCP transport did not work properly (welcome message size slightly wrong, by itself mostly harmless)*
    * some plaintext messages were not properly encapsulated by the sender (resulting in failure to connect)*
    * configuration file generated by gnunetd if no file was found was outdated (contrib/gnunet.root was ok, though)*


Offline bitz

  • Advanced
  • ***
  • Posts: 160
  • Karma: +2/-0
  • Planet Peer Community
    • View Profile
    • My Blog
Re: GNUnet 0.7.0pre2 released
« Reply #1 on: April 05, 2005, 11:23:02 AM »
Perhaps someone could be so kind as to sum up gnunet's current status?

Offline crypton

  • Elite
  • *****
  • Posts: 1699
  • Karma: +10/-0
    • View Profile
Re: GNUnet 0.7.0pre2 released
« Reply #2 on: April 05, 2005, 11:32:50 AM »
Also, some features have already been deferred for later in the 0.7.x series:*

    * gnunet-chat*
    * DHT
    * meta-data finder*
    * directory browser*
    * various optimizations*

The goals for this pre-release are to:*

    * bootstrap the network*
    * see if there are any problems with the code if the size of the network goes beyond the small testbeds that individual developers can run*
    * give packagers a chance to test their build-process*
    * give developers an opportunity to test (graphical) user interfaces*

In summary, please try to run this code, even if it maybe not super-stable and lacks a graphical user interface.* Please report any bugs you find to mantis.* Note that all protocols and storage formats have changed, so you may want to make sure that when trying 0.7.0 you do not accidentially overwrite or use files from 0.6.x.* Migrating data from 0.6.x is not possible.* However, you can run 0.7.x and 0.6.x in parallel if you use different ports and directories.*

these are the highlights:*

    * highly modular and pluggable architecture*
    * extensive test-suite for testing individual modules*
    * improved session key exchange protocol*
    * support for peer-to-peer messages up to 64k*
    * remote procedure call API*
    * distributed hash table API*
    * simple but powerful APIs for writing GNUnet user interfaces*
    * clear separation of anonymous routing and file-sharing concerns*
    * new meta-data encoding allows arbitrary meta-data including thumbnails*
    * new database abstraction for improved performance (only MySQL and SQLite supported)*
    * new ECRS encoding with KBlocks for improved security against router-in-the-middle data corruption attacks*
    * content expiration (after originator dies) without breaking properties of the GNUnet anonymity protocol*
    * different anonymity requirements can be specified for each shared file*
    * improved internationalization and localization*

This was achieved by doing an almost complete re-write of the code.* The list is incomplete in that for the final 0.7.0 release additional changes are planned.* In particular various aspects of how content and peer-to-peer messages are encoded will change (a bit more).* In consequence, this release:*

    * breaks compatibility with all previous versions*
    * will also not be compatible with any future version*
    * does not feature a graphical user interface*
    * will not bootstrap (as in, there is no hostlist to find other peers)*
    * has terrible performance (intentionally for debugging)*
    * has known critical bugs*

So why release the code at this stage?* The answer is, this is a pre-release with the goal to:*

    * give packagers a chance to update their build-process*
    * give developers an opportunity to create (graphical) user interfaces based on the new APIs*
    * get some early feedback about critical problems*
    * find portability problems for platforms that we do not have access to
« Last Edit: April 05, 2005, 11:38:18 AM by defnax »