Skip to content


Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
Fetching contributors…

Cannot retrieve contributors at this time

67 lines (47 sloc) 2.433 kb
18 October 2009: dht-0.10
* Send nodes even when sending values. This is a violation of the
protocol, but I have been assured that it doesn't break any deployed
implementation. This is also what both libtorrent and uTorrent do.
* Give up immediately on a search peer when no token was provided. This
is a very reasonable extension to the protocol, and certainly doesn't
break anything.
* Parse heterogeneous values lists correctly. This is mandated by BEP 32.
20 September 2009: dht-0.9
* Fixed incorrect computation of number of nodes.
* Made the initial bucket split eagerly (speeds up bootstrapping).
* Fixed initial filling of search buckets (speeds up searches).
28 July 2009: dht-0.8
* Fixed a crash when expiring the first search on the list.
* Fixed freeing of the search list when uniniting with dofree = 1.
24 June 2009: dht-0.7
* Removed the fixed limit on the number of concurrent searches, we now
use a linked list.
* Fixed build on FreeBSD (thanks to Humihara and Charles Kerr).
22 May 2009: dht-0.6
* Fixed a buffer overflow (when reading) in parse_message.
* Fixed slightly inacurrate metric computation when searching.
* Removed a slightly inaccurate shortcut when responding to find_nodes.
* Relaxed the rate-limiting parameters to 4 requests per second.
19 May 2009: dht-0.5
* Made reading of /dev/urandom a function provided by the user.
* Implemented the ``v'' extension that identifies node implementations.
18 May 2009: dht-0.4
* Fixed the handling of tokens in announce_peer messages.
* Implemented backtracking during search when nodes turn out to be dead.
17 May 2009: dht-0.3
* Fixed a number of incorrectly formatted messages.
* Changed reply to find_peers to spread the load more uniformly.
* Fixed a bug that could cause premature splitting.
* Implemented rate limiting.
* Changed some time constants to be less chatty.
* When determining if a bucket is fresh enough, we now only take replies
into account.
* dht_get_nodes now returns nodes starting with our own bucket.
* Tweaked the memory allocation strategy for stored peers.
17 May 2009: dht-0.2
* Fixed a crash in dht_uninit.
* Added support for saving the list of known-good nodes.
* Changed the interface of dht_nodes to provide the number of nodes that
recently sent incoming requests.
13 May 2009: dht-0.1
* Initial public release.
Jump to Line
Something went wrong with that request. Please try again.