- When socket is closed, set sessionID and LS to null,
close subsession and set its sessionID and LS to null
- Checks on client side for null session ID
- Check for null session in Destroy Session message
- Don't kill I2CP connection due to a bad session ID
in a SendMessage, just drop the message and send
a MessageStatusMessage
- Log tweaks
Catch null SessionId in messages and stop connection instead of NPE
Wait for LS in SubSession in connect() so we don't send data w/o
a session ID and leaseset
Log tweaks
isn't destroyed and immediately replaced by i2ptunnel,
which caused dup shared clients in a race at startup;
Clarify session exception text if not open
isn't treated as CLOSED and immediately replaced by i2ptunnel,
which caused dup shared clients in a race at startup
(possible related tickets #642, #650, #815, #1545)
Start availability notifier in subsession
Availability notifier cleanup
Various log tweaks added while chasing this down
Better subsession state management
I2PSocketManagerFull verifies subsession to force connect()
Successfully tested
Fix sending CreateSessionMessage for subsession
New AliasedTunnelPool for subsessions, don't reuse TunnelPool,
so it has its own settings
Fix addAlias()
Simplify refreshSettings()
Send status message on subsession create failure
Fix settings for subsession
Fix NPE in requestLeaseSet()
Fix setting new session ID in SessionStatusMessage
Fix subsession support detection
Streaming: one socket manager, multiple connection managers.
Change data structure for subessions in socket manager
Subsession cleanup on destroy
I2PTunnel: add DSA subsession for non-DSA shared client
Javadocs
Work in progress.
Router-side I2CP mostly done.
Client-side I2CP mostly done but undecided on how to handle
listeners.
Streaming stubbed out but may be wrong, may need multiple socket managers,
not clear how to proceed.
I2PTunnel not started.
Blacklist of DSA-only dests not started.
Router leaseset publishing not correct. Not clear whether to have
additional tunnel pools with flags, or put the tunnel pools into
the client hashmap twice. Client config contains destination,
may need to move that to tunnel pool.
- Redefine the repliable datagram signature for non-DSA_SHA1 sig types;
was the sig of the SHA-256 of the payload, now the sig of the payload itself.
This is an incompatible change but nobody is yet using the new
sig types for datagram applications.
- Don't pollute the hash cache with hashes of payloads
- Check for too-big datagrams
- Remove assertion check
- Cleanups
