Update of /cvsroot/freenet/freenet
In directory sc8-pr-cvs1:/tmp/cvs-serv16753
Modified Files:
TODO
Log Message:
6224:
More debugging of PeerHandler and supporting microarchitectural changes.
- Splitfile fetch between 2 test nodes works again, rather than failing with RNFs.
- Fix totalDataSent i.e. fix OCM's report of total data sent on a conn
- Encrypt data at send time. Not in advance in packet.
- Locking changes.
- Add hasTrailer() to Message. Implement it on all Messages.
- Make a distinction between needsConnection as in can we route to it, or do we want a
new connection if possible.
- Keep separate queues for messages with and without trailers in PeerHandler.
- Messages without trailers take priority.
- needConnection(true) returns false if we have no messages queued and no trailers
sending and we are not in the RT.
- Remove closed conns from the PeerHandler when we find them. They will still be in
the OCM.
- SendData: Count size of padding sent the same way we count the non-padding - don't
add it on and therefore don't finish until after we have sent it.
Add a TODO about the data: protocol
- Cancel outwardSender callback in Pending after we get Accepted. We may have a bug
causing notifications to be lost.
- Fix possible NPE in FnpLink.encryptBytes.
Logging
Also some fixes and optimizations since 6223 by other people.
Index: TODO
===================================================================
RCS file: /cvsroot/freenet/freenet/TODO,v
retrieving revision 1.27
retrieving revision 1.28
diff -u -r1.27 -r1.28
--- TODO 30 Aug 2003 23:16:50 -0000 1.27
+++ TODO 8 Oct 2003 22:54:18 -0000 1.28
@@ -33,4 +33,13 @@
+ Move bookkeeping variables somewhere else - RequestState base class, or a large
object pointed to by Node or Core. Not _in_ Core, where they are now. Consider using
logging mechanism to collect info.
+ Allow multiple addresses in bindAddress - should be easy enough now that we have nio
+ Look into the possibility of supporting _SAFE_ (this is hard, we need to prevent it
from knowing what is in the datastore even by timing attacks) scripting in fproxy
-+ Look into including the length of the data in the key for a CHK. Useful for apps
using prefetch - which are inevitable.
\ No newline at end of file
++ Look into including the length of the data in the key for a CHK. Useful for apps
using prefetch - which are inevitable.
++ <Fruser_> data:image/gif;base64,R0lGODdhIAAgAIAAAAAAAPj8+CwAAAAAIAAgAAAClYyPqcu
+ 9AJyCjtIKc5w5xP14xgeO2tlY3nWcajmZZdeJcGKxrmimms1KMTa1Wg8UROx4MNUq1Hry
+ cMjHT9b6xKxaFLM6VRKzI+pKS9XtXpcbdun6uWVxJXA8pNPkdkkxhxc21LZHFOgD2KMoQ
+ Xa2KMWIJtnE2KizVUkYJVZZ1nczBxXlFopZBtoJ2diXGdNUymmJdFMAADs=
+ ( works in Mozilla at least )
+ - http://www.ietf.org/rfc/rfc2397.txt
+ -> inline images - do we want to support this in fproxy? We would need to
+ check the MIME type and either block or filter dangerous content.
+
_______________________________________________
cvs mailing list
[EMAIL PROTECTED]
http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/cvs
