7e08250f8c
refactor setting successors and predecessors
2020-06-17 14:29:30 +02:00
2269357ed0
deleting RingMap entries, list conversion
2020-06-17 02:21:37 +02:00
6142ee61d7
WIP: implement adding, setting and taking RingMap entries.
...
contributes to #48
2020-06-16 23:51:53 +02:00
22a6becf6b
fix all previously working tests
2020-06-15 16:41:11 +02:00
061bce2b08
adjust types to refactored RingMap NodeCache
2020-06-15 15:14:11 +02:00
6a98b5c6da
fix RingMap function types
2020-06-15 13:53:06 +02:00
b179357ab0
generalise NodeCache implementation to make it usable for neighbour nodes as well
...
contributes to #48
2020-06-13 21:41:32 +02:00
7612f5532a
create a test for parts of #48
2020-06-13 14:56:05 +02:00
2739b47162
send stabilise to certain successor or predecessor
2020-06-12 15:49:12 +02:00
1204457a2a
make sure that predecessors are smaller and successors are larger than node ID
2020-06-10 16:53:17 +02:00
2c98d8507d
implement stabilise request sending and parsing
...
contributes to #44
2020-06-09 22:11:42 +02:00
f15d83baff
Refactor predecessors and successors to hold RemoteNodeStates
...
- neighbour nodes need to be contacted reliably
- Only holding NodeIDs requires a cache lookup for getting hostname and
port. This is brittle as the entry could've been purged from cache.
- refactored ASN.1 scheme, types and add/ sort/ remove implementations
closes #46
2020-06-09 15:21:22 +02:00
67cba1b69b
fixup! rename responseTo -> requestID to avoid partial record accessors
2020-06-08 23:18:12 +02:00
4e359775ec
add some debug output prints
2020-06-08 20:31:22 +02:00
2c827ea326
rename responseTo -> requestID to avoid partial record accessors
2020-06-08 20:31:18 +02:00
3892dc91aa
Merge branch 'mainline' into dhtNetworking
2020-06-08 18:58:23 +02:00
7c87a578d3
move Main executable to app/ closes #18
2020-06-08 18:57:53 +02:00
6699237243
respond to and handle QueryID requests
...
closes #28
2020-06-08 00:36:59 +02:00
eec751584c
make setPredecessors and setSuccessors enforce the ordering + size limit
...
closes #47
2020-06-07 23:45:28 +02:00
0a9b0547c6
forgot handling the successors and predecessors of the leaving node
...
contributes to #28
2020-06-07 00:21:14 +02:00
e00da9b84f
respond to and handle Leave requests
...
contributes to #28
2020-06-06 18:02:55 +02:00
53308080db
respond to Stabilise requests
2020-06-06 17:31:20 +02:00
43d72128d2
respond to Ping requests
2020-06-06 17:05:54 +02:00
cb769e088f
add test for serialiseMessage part numbering
2020-06-06 13:45:35 +02:00
e32f0c9146
process and respond to join requests
...
- serialiseMessage now starts numbering parts from the first number it
gets, to be able to continue responding after having ACKed previous
parts
contributes to #28
2020-06-06 13:45:18 +02:00
914e07a412
change NodeCache protection to STM
...
- putting the NodeCache behind an IORef had been chose because those
could've been read non-blocking
- the same is true for TVars. The performance characteristics are likely
worse, but at the advantage of composability within STM monads
2020-06-05 22:07:47 +02:00
dc2e399d64
protect concurrent node state access with STM
...
- for allowing concurrent access to predecessors and successors, the
whole LocalNodeState is passed wrapped into an STM TVar
- this allows keeping the tests for the mostly pure data type, compared
to protecting only the successor and predecessor list
contributes to #28
2020-06-04 22:31:41 +02:00
f42dfb2137
dispatch incoming requests to their response functions
...
- contributes to #28
2020-06-04 14:48:05 +02:00
0660bce299
acknowledge parts when receiving partial requests
2020-05-31 23:21:55 +02:00
88104de9bf
periodically purge request parts
2020-05-31 21:07:40 +02:00
bcd1c34c7c
manage incoming request parts before handling
2020-05-31 01:14:09 +02:00
2542091379
adjust rest of code to new message structure
2020-05-30 13:52:06 +02:00
f8d444d5b6
FediChordMessage: last part has falg instead of parts number in each msg
...
Motivation: Including the number of parts in each message part requires
the total number of parts to be known in advance, making dynamic
responses based on the received data difficult
2020-05-30 13:07:28 +02:00
fea9660f80
WIP: start implementing incoming request handling
2020-05-30 01:10:56 +02:00
96e61b726f
adjust hlint hints to ignore lambda-case
2020-05-29 21:30:18 +02:00
61818c58a9
main server thread structure
2020-05-29 17:39:35 +02:00
b4ecf8b0aa
catch and handle more join errors
2020-05-27 23:37:32 +02:00
6ff765c63e
catch and handle bootstrap join errors
2020-05-27 18:59:38 +02:00
27e5c5f9ce
bracket all socket-using operations to avoid resource leakage
2020-05-27 18:08:46 +02:00
b1c5c5e5f4
try all bootstrap nodes until successfully joined
2020-05-26 20:54:02 +02:00
702684b1a9
split fediChordJoin into general purpose and bootstrapping part
2020-05-26 11:02:29 +02:00
43eb04dfea
preliminary passing of bootstrap nodes in Main to fediChordJoin
2020-05-26 09:39:02 +02:00
ad1465c5fe
use global cache adding function
2020-05-26 08:55:44 +02:00
e3bfa26ddb
join request + large FediChord refactoring
...
- implement sending of initial join request sending, response parsing
and cache population (untested but compiles)
- refactor basic types and their functions into Hash2Pub.FediChordTypes
to prevent import loops, leaving Hash2Pub.FediChord to contain the
high level actions called from Main
2020-05-25 22:03:24 +02:00
d5841d13fd
extract queryID message sending function so it can be used for first join
2020-05-22 22:12:38 +02:00
16769d1395
Merge branch 'refactorNodeState' into dhtNetworking
...
closes #20
2020-05-22 00:06:10 +02:00
bbe7078369
adjust tests to work with new NodeState types
...
closes #20
2020-05-22 00:05:23 +02:00
e8091b0a29
change all function definitions to new NodeState types
...
- adjust implementation
contributes to #20
2020-05-21 23:48:36 +02:00
fe673dc255
make NodeState a typeclasse, define setters and getters on its representations
...
contributes to #20
2020-05-21 23:40:10 +02:00
99a2b0ba09
add instance deriving instructions
2020-05-20 19:27:51 +02:00
