The default service path in Windows is fugly and not very convenient. I2P uses
the correct path, but if you want to access snark or eepsite data, one must go to
%SYSTEMROOT%\config\systemprofile\AppData\Roaming\I2P\ (Vista/7) or
%SYSTEMROOT%\system32\config\systemprofile\Application Data\I2P (XP/2003). If
this wasn't bad enough, in some cases one must take ownership of this path and
grant permission to him- or herself to access the folder.
With this changeset, I'm setting the path to %ALLUSERSPROFILE%\Application
Data\I2P as well as adding a shortcut to the I2P folder in the Start menu.
- Replace the old parallel lookup method with a true
Kademlia lookup that iteratively queries additional floodfill
peers returned in DatabaseSearchReplyMessages. This is a more
efficient and reliable lookup that will work much better
when not all floodfill peers are known, and it removes a serious
limitation to network growth.
- Limit max number of DSRM entries to look up
- Cleanups, javadocs, log tweaks
- Update Italian, Chinese, Spanish, Polish, Swedish,
and Vietnamese translations from transifex
** po files contain errors, to be fixed in next checkin
- Allow NTP to reinitialize clock after the comm system
in the first minute of uptime
- Fix i2ptunnels not restarting
- Increase minimum forward clock shift for soft restart
- Reduce minimum backward clock shift for soft restart
- Signal the I2CP client with a different message when restarting
- I2CP client reconnects when receiving restart message
- Follow all DSRM's, not just the last one, by moving the code
from the match job to the selector
- Update peer profile after SingleSearchJob
- Cleanups, javadocs, log tweaks, final
- Explore even more aggressively at startup
- Increase SingleSearchJob timeout
- Increase time at startup for processing DSRM
- Cleanups, final, javadoc, concurrent
- Revert change in UDPPacket that caused massive corruption
- Reduce buffer sizes from 2048 to 1536
- Discard too-large received packets sooner
- More cleanups
- Notes on MTU
- Complete rewrite of OutboundMessageFragments for
concurrent and for efficiency to avoid O(n**2) behavior
- Queue a new send immediately after a packet is acked
- Cleanups, log tweaks, javadocs, final
