Details of this build:
Various bug fixes, some of them pretty serious. Comes to a fairly small
amount of actual code though. Lots more changes are being tested and debugged on unstable.
* Rate limiting was almost completely broken due to a stupid bug. Fixed
it.
* Major routing fix: many of our running averages were, for various
reasons, at 0.999999999999 or so. With the current running average
implementation for probabilities, the result of this is that the
running average will never change (because the closer it is to 0.0 or
1.0, the less sensitive it is). We were supposed to be doing bounds
checking to ensure that this wasn't a problem; there was a bug in the
bounds checking :(. The result of this bug was that some of the major
factors in the NGRouting calculation were very often completely wrong!
The current implementation is suboptimal, there is a replacement being tested on unstable, but the code merged in this build should be
much better than the previous builds!
How could that happen???
It is the same problem you had in the unstable branch, isn't it?
So why did the modification, which have caused the problem in unstable, get into the code of the stable branch, before it was tested for a longer period (at least a day or two) in the unstable network? For what are you running the unstable network?
I DO NOT UNDERSTAND IT!
* Fixed a NullPointerException that would cause the web interface to stop working after a while. * Show some more useful detail next to probability running averages on the per node pages linked from the routing table page. * Don't leak a temp file when an insert fails. * A minor optimization in DSAPublicKey.equals(), and some minor code style changes.
Regards Rudi
-- Rudolf Krist PGP Key Id: 0x5C1708B8
_______________________________________________ Support mailing list [EMAIL PROTECTED] http://news.gmane.org/gmane.network.freenet.support Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support Or mailto:[EMAIL PROTECTED]
