1 files changed, 38 insertions, 83 deletions

diff --git a/ISSUES b/ISSUES
index 2a13bdf..6ab87fb 100644
--- a/ISSUES
+++ b/ISSUES
@@ -1,91 +1,46 @@
-== testbed ==
-
-padding: PACKED / NETWORK_STRUCT don't really pack some misaligned structs:
-  struct GNUNET_TESTBED_PeerConfigurationInformationMessage
-is misaligned anyway, *but* sizeof the struct is 84, when it should be 82!
-
-see https://gnunet.org/bugs/view.php?id=3043
-
-I managed to break testbed (this is a GNUNET_assert, no GNUNET_break)
-  Sep 24 20:06:14-146897 testbed-18586 ERROR Assertion failed at testing.c:1548.
-but this only occurs *after* the shutdown and when trying to reconfigure a peer!
-
-OPTIONS/PREFIX does not work with testbed helper
-
-what works (and is tested) now in org.gnunet.testbed:
- * creating and destroying peers
- * starting and stopping peers
- * getting peer id / configuration
- * overlay-connecting two peers
-other stuff:
- * does gnunet-java need barriers?
- * what's the state of the overlay implementation
- * does service_connect do anything other than retrieve the configuration?
-  * primary purpose is to be able to have opqueue management, right?
- * why do we have manage service when we have arm?
-  * someone has to start arm, I suppose?
-
-== crypto ==
-two types of public keys, do I understand the reasons correctly:
- * NaCl compatibility
- * fastest representation for each type of operation
-
-why is there no way to convert from signing key -> encryption key?
- * other way around is not possible
-
-
-== voting ==
+testbed api is now quite complete
+ * problem with reconfigure was a bug in the testbed service
+ * only thing left to to is to implement arm for manageService
+  * (BUT: I don't get how arm works in conjunction with testbed -- when the arm api can't connect to
+    arm it tries to start the arm service *locally*, which is often wrong ...)
+  * talked to sriharsha about it
+
+gnunet-java finally has a proper stream tokenizer ...
+ * and locking has been removes from Scheduler, Helper is properly implemented
+
+new java API implementation for identity :)
+ * why is there no way to import keys?
+  * at least API-wise
+ * voting should use identity ...
+
+timing information in ballot: we talked about round times,
+but shouldn't this be done transparently by consensus?
+aren't START_TIME, CLOSING_TIME, QUERY_TIME, END_TIME and maybe CONSENSUS_PARAMS be enough?
+
+certificate authority: is the CAs pubkey different from its peer?
+ * it definitely makes sense to have it seperate for other cert methods
+  * like sending a cert per email, snail mail, social network ...
+ * how do we specify how to reach the cert? one peer? multiple peers?
+  * maybe use GADS? (I know very little about GADS right now)
+ * does it even make sense to have the CA service at this time?
 
 what now works:
- * creating a ballot and registering it
- * selecting a choice in the ballot and submitting it
-  * vote + 512-bit ballot GUID
- * querying the results (also with the GUID)
-
-next steps??
-- Ballot: (a) voter ID, (b) election information, (c) election eligibility group (Bavaria 18+), (d) ballot GUID
-- no double submission: check: no duplicate voter ID (for any given election)
-- no early queries
-- voter <--> CA interaction
-  => ideally, CA has a list of voter public keys or at least a policy (i.e. "accept all") 
-     for a certain voter group (i.e. Bavaria 18+). 
-  => CA also has a private CA key (surprise!)
-  => Voter has voter public key and voter group, sends THOSE to CA;
-  => CA signs: this voter is in this voter group (with expiration time for signature!)
-  => voter takes signature, can re-use it for ALL elections with this voter group
-- Authorities have CA public keys, check voter group sigs.
-
-what should be stored in the ballot about authorities: peer identities or peer public keys?
-=> public keys
-
- * pubkey makes more sense, but I have the feeling that GNUnet prefers to use
-   peer IDs whenever possible
-=> Nah.
-
-docs of identity service say:
-  For giving names to other users and manage their public keys securely, we use GNS.
-In what way does GNS manage public keys? 
-  GNS allows you to give names to other user's public keys.
-Does GNS in any way relate peer identities to public keys?
-  No.
-
-persistent storage of tallies?
-  => Later.
+ * authorities check for double vote submission, late / early voting, early result query,
+   double registering, vouchers (I don't like to call vouchers receipts because that term has
+   a certain meaning in voting literature already)
 
-now that the pubkey is smaller, are there reasons for still having
-a separate peer identity?
- * except for that it would be a lot of work to change now
-=> exactly.
+but: group certs are not checked right now
 
-the command line took looks a bit complicated now, any suggestions?
-=> Later.
+how to test voting without waiting too long or missing the right time?
 
-asking permission to vote should not require the private key, right?
- * how do we do this interface-wise? pubkey on the command line?
-=> Identity service, ego!
+any progress on crypto?
+ * luckily the private key has the same size, so identity can be used
 
-== general java stuff ==
-what's your opinion on option types: should they be used more in gnunet-java?
+next:
+ * use identity
+ * multiple authorities should do consensus on unencrypted votes
+ * tests for this with testbed
+ * encrypting votes
 
-We have a lot of callbacks in java that all do almost the same. Should we have a generic callback type?
+suggestions?