95bf68f3f5
Correct some strange naming
2018-05-11 16:24:18 -06:00
bcdb058492
cargo fmt
2018-05-11 13:06:05 -06:00
3d77fa5fbc
Merge branch 'master' into 153-panic-cleanup
2018-05-11 11:40:20 -07:00
250830ade9
cargo fmt run
2018-05-11 11:38:52 -07:00
a49e664e63
Merge branch '156__remove_user_keys_in_mintdemo' of github.com:rlkelly/solana into 156__remove_user_keys_in_mintdemo
2018-05-11 14:07:48 -04:00
f20380d6b4
changed RwLock to RefCell
2018-05-11 14:07:41 -04:00
05a5e551d6
Merge branch 'master' into 156__remove_user_keys_in_mintdemo
2018-05-11 13:00:44 -04:00
d278b71cb2
added tests and utility method for key generation
2018-05-11 12:55:05 -04:00
8b7f7f1088
Generalize next tick functions to carry events
2018-05-11 09:45:42 -06:00
4f629dd982
Add events socket instead of modifying the existing socket
2018-05-10 16:54:43 -06:00
a80991f2b3
Fixes for serializing entries over blobs and reorg into ledger
2018-05-10 15:30:30 -07:00
7ce11b5d1c
Cleanup: use full words for field names
...
and optionally for variable names
2018-05-09 16:19:42 -06:00
f2d4799491
Cleanup: field names should be nouns
2018-05-09 16:14:40 -06:00
ded28c705f
Tuck away the Historian
...
The Historian is now just a utility of the accounting stage.
2018-05-09 12:25:19 -06:00
6967cf7f86
Boot sync_channel()
...
This is less useful now that we send Vec<Event> instead of Event.
2018-05-09 11:43:16 -06:00
52f1a02938
Delete historical artifact
...
This was just to explain Proof of History. We have better explanations
elsewhere. Delete!
2018-05-09 07:53:24 -06:00
40c50aef50
deterministic random wallet generationg
2018-05-09 00:07:19 -04:00
1dca17fdb4
cargo +nightly fmt
2018-05-08 18:59:01 -06:00
785e971698
AccountantSkel -> Tpu
...
The terms Stub and Skel come from OMG IDL and only made sense while
the Stub was acting as an RPC client for the the Accountant object.
Nowadays, the Stub interface looks nothing like the Accountant and
meanwhile we've recognized the multithreaded implementation is more
reminiscent of a pipelined CPU. Thus, we finally bite the bullet and
rename our modules.
AccountantSkel -> Tpu
AccountantStub -> ThinClient
Up next will be moving much of the TPU code into separate modules,
each representing a stage of the pipeline. The interface of each
will follow the precedent set by the Historian object.
2018-05-08 17:40:02 -06:00
2bfa20ff85
Merge pull request #182 from garious/split-request
...
Control port prep
2018-05-08 17:11:34 -06:00
474a9af78d
Merge pull request #187 from sakridge/fix_blob_size_check
...
Trust the recorder not to give us more than we can serialize
2018-05-08 17:11:18 -06:00
4870def1fb
Fix default client port, server uses 8000-8002 for gossip.
2018-05-08 15:40:55 -07:00
3e73fb9233
Trust the recorder not to give us more than we can serialize
...
Also run client for 10 seconds, 5 is bit too short
2018-05-08 15:23:41 -07:00
9ff1a6f0cd
Add a thread to support thin clients
2018-05-07 21:44:44 -06:00
893011c3ba
Process events instead of processing only transactions
...
Prep work to allow clients to send any type that can end up in
the ledger.
2018-05-07 14:51:13 -06:00
e8f5fb35ac
Multinode fixes and test
...
* Replace magic numbers for 64k event size
* Fix gossip, dont ping yourself
* Retransmit only to listening nodes
* Multinode test in stub marked unstable
2018-05-06 21:36:06 -07:00
de680c2a8e
Remove duplicate state
2018-05-03 13:24:37 -06:00
c2e2960bf7
Add broadcast impl
2018-05-03 10:34:01 -07:00
385d2a580c
Merge pull request #168 from aeyakovenko/fix_multi_host_client_demo
...
multi host client demo
2018-05-03 10:21:41 -06:00
b65f04d500
multi host client demo
...
Bind to the same interface as the user supplied client address.
2018-05-03 08:28:11 -07:00
4b9f115586
Hoist Historian input
2018-05-02 16:35:37 -06:00
c5cc91443e
Rename sender/receiver to input/output
2018-05-02 15:54:53 -06:00
d415b17146
sleepless demo to complement sleepless nights
...
18 ktps on macbook pro, no gpu
2018-04-26 13:17:38 -06:00
9ed953e8c3
Fix rebase fails
2018-04-26 09:35:10 -06:00
b60a98bd6e
Startup log can reference IDs without itself
2018-04-26 08:42:34 -06:00
a15e30d4b3
Report transactions processed
2018-04-26 08:42:34 -06:00
d5d133353f
Port blocking stub functions to new stateful ones
2018-04-26 08:42:34 -06:00
58860ed19f
WIP: New demo that makes better use of the parallelized accountant
2018-04-26 08:42:34 -06:00
583f652197
Generate genesis log for the demo
...
This log contains a bunch of transactions that generate new
accounts, so that transactions to and from them can be processed
in parallel.
2018-04-26 08:42:34 -06:00
38fdd17067
Add initializing log message to server
...
Handy when gesesis block is large.
2018-04-26 08:42:34 -06:00
807ccd15ba
Add solana-mint-demo CLI
...
This extends solana-mint with additional data that will be used by
both solana-client-demo and creating the demo's genesis block.
2018-04-26 08:42:34 -06:00
34d1805b54
Add FutureResult to return a Future that immediately resolves
2018-04-25 19:23:24 -07:00
3da1fa4d88
improve the error messages
2018-04-21 21:52:55 +08:00
8181bc591b
Add -h/--help options for client-demo and testnode
2018-04-19 10:22:31 -07:00
60015aee04
report serde parse errors to stderr
2018-04-19 23:51:57 +08:00
b91f6bcbff
report parse errors to stderr
2018-04-19 22:24:46 +08:00
0a30bd74c1
Tell verifiers when not to parallelize accounting
...
Without this patch, many batches of transactions could be tossed
into a single entry, but the parallelized accountant can only
guarentee the transactions in the batch can be processed in
parallel.
This patch signals the historian to generate a new Entry after
each batch. Validators must maintain sequential consistency
across Entries.
2018-04-12 21:08:53 -06:00
3b9ef5ccab
Fix the nightly build
2018-04-11 20:24:14 -06:00
c934a30f66
commandline options for client and testnode
2018-04-09 21:14:52 -07:00
f4466c8c0a
Change for cuda verify integration
2018-04-05 20:00:44 -07:00
0c60fdd2ce
Make accountant thread-safe
...
Before this change, parallel transaction processing required locking
the full accountant. Since we only call one method,
process_verified_transaction, the global lock equates to doing no
parallelization at all. With this change, we only lock the data that's
being written to.
2018-04-04 12:33:03 -06:00
5ac7df17f9
Implement window service
...
Batch out of order blobs until we have a contigious window.
2018-04-03 13:53:19 -07:00
c323bd3c87
Fix clippy warnings
2018-04-03 09:55:33 -06:00
94eea3abec
fmt
2018-04-02 21:15:21 -06:00
daadae7987
Move replaying ledger out of accountant
2018-04-02 14:51:55 -06:00
2b788d06b7
Move the historian up to accountant_skel
2018-04-02 14:41:07 -06:00
da2b4962a9
Move verify_slice() into a trait
2018-04-02 11:43:38 -06:00
46e8c09bd8
Revoke API access to first_id
2018-04-02 09:30:10 -06:00
2f1e585446
Better benchmark
...
Tolerates dropped UDP packets
2018-03-29 13:41:11 -06:00
878ca8c5c5
Add microbenchmark for signature verification
2018-03-28 22:02:47 -06:00
0bec360a31
Revert TCP sync of ledger
...
The feature was too rushed. We technically don't need it until we
implement consensus. It'll come back another day (with many more tests!)
2018-03-28 20:16:15 -06:00
98c0a2af87
tx confirmed/sec ---> tx processed/sec
...
Before this patch, we were waiting until the full log was
sent back across the wire, parsed, and interpreted. That was giving
us a metric of "transactions confirmed per second" instead of
"transactions processed per second". Instead, we'll just send one
tiny packet back with the balance. As soon as the balance is what
we expect it to be, we end the benchmark.
2018-03-28 16:51:21 -06:00
849bced602
Fix up client demo
2018-03-28 14:40:58 -06:00
27f29019ef
Merge pull request #83 from garious/tcp-client
...
TCP subscription service
2018-03-28 13:19:38 -06:00
116166f62d
Rename project: silk -> solana
2018-03-27 16:25:12 -06:00
13206e4976
Let clients subscribe to the ledger over TCP
...
TODO: Add more tests
Fixes #27
2018-03-27 14:46:24 -06:00
ddb21d151d
Nightly rustfmt
...
Format code with the nightly version of rustfmt, which sorts imports.
2018-03-26 22:03:28 -06:00
14239e584f
fix writer
2018-03-26 21:36:29 -06:00
9a437f0d38
Revive silk-testnode
2018-03-23 21:49:28 -06:00
383d445ba1
Clippy review
2018-03-22 14:15:29 -06:00
803dcb0800
Mutex<bool> -> AtomicBool
2018-03-22 14:05:23 -06:00
9f232bac58
Allow clients to sync the ledger
...
Fixes #4
2018-03-21 15:46:49 -06:00
64af37e0cd
logger -> recorder
...
Free up namespace for a traditional runtime logger.
2018-03-19 10:16:21 -06:00
9f9b79f30b
log -> ledger
...
Free up namespace for traditional runtime logs.
2018-03-19 10:09:19 -06:00
eb94613d7d
Use streaming socket interface within accountant
...
Pull messages from streamer process them and forward them to the sender.
2018-03-11 23:41:09 -05:00
49281b24e5
Move Tick out of Event
...
Every Entry is now a Tick and the entries contain events.
2018-03-09 17:22:17 -07:00
cc9f0788aa
Batch events
...
It's now a Tick that locks down event order. Before this change, the
event order would be locked down in the order the server sees it.
Fixes #59
Fixes #61
2018-03-09 16:16:33 -07:00
5dd567deef
Rename Genesis to Mint
...
Genesis is a story of creation. We should only use that term to
for the event log that bootstraps the system.
2018-03-07 17:08:15 -07:00
b6d8f737ca
Introducing, the mint
...
Use the mint to pair a new private key with new tokens.
2018-03-07 16:58:04 -07:00
491ba9da84
Add accessors to keypairs and signatures
2018-03-07 15:32:22 -07:00
a420a9293f
Fix demo
2018-03-07 11:37:30 -07:00
9834c251d0
Boot genesis block helper
...
Before this change, if you wanted to use a new Transaction
feature in the genesis block, you'd need to extend its
Creator object and associated methods. With yesterday's
addtions to Transcation, it's now so easy to work with
Transactions directly that we can get rid of the middleman.
Also added a KeyPair type alias, so that ring could be easily swapped
out with a competing library, if needed.
2018-03-07 11:10:15 -07:00
d1b6206858
Monomorphisize Entry and Event
...
Transaction turned out to be the only struct worth making generic.
2018-03-06 20:29:18 -07:00
b725fdb093
Sha256Hash -> Hash
...
Because in Loom, there's just the one. Hopefully no worries that it
shares a name with std::Hash.
2018-03-06 17:40:01 -07:00
1436bb1ff2
Move entry into its own module
...
Hmm, Logger doesn't depend on log.
2018-03-06 17:40:01 -07:00
5a44c36b1f
Move hash into its own module
2018-03-06 17:40:01 -07:00
f09952f3d7
DRY up transaction signing
...
Cleanup the big mess I copy-pasted myself into.
2018-03-06 16:34:25 -07:00
4376032e3a
data -> asset
...
'data' is too vague.
2018-03-06 14:50:32 -07:00
624c151ca2
Add signature module
...
Because things other than transactions can be signed.
2018-03-06 12:48:28 -07:00
b019416518
Move verify into methods
...
A little overly-coupled to Serialize, but makes the code a lot tighter
2018-03-06 12:27:08 -07:00
4fcd9e3bd6
Give Transaction its own module
2018-03-06 12:18:17 -07:00
66bf889c39
Rename Transfer to Transaction
...
struct names should be nouns
2018-03-06 11:54:47 -07:00
6543497c17
Move Transaction data into its own struct
...
This will allow us to add addition transfer types to the log.
2018-03-06 10:50:32 -07:00
720c54a5bb
Allow balances to be negative
...
* Will allow owners to loan token to others.
* Will allow for parallel verification of balances without spilling
over 64 bits.
Fixes #43
2018-03-05 17:30:53 -07:00
a4336a39d6
Initialize the testnode from a log
...
$ cargo run --bin silk-genesis-file-demo > demo-genesis.json
$ cat demo-genesis.json | cargo run --bin silk-genesis-block > demo-genesis.log
$ cat demo-genesis.log | cargo run --bin silk-testnode
2018-03-05 15:34:44 -07:00
298989c4b9
Generate log from Genesis
2018-03-05 13:03:56 -07:00
48c28c2267
Transactions now require a hash of the last entry they've seen
...
This ensures the transaction cannot be processed on a chain
that forked before that ID. It will also provide a basis for
expiration constraints. A client may want their transaction
to expire, and the generators may want to reject transactions
that have been floating in the ether for years.
2018-03-05 12:48:14 -07:00
d76ecbc9c9
Don't block the server
2018-03-05 11:39:59 -07:00
955f76baab
Finally, genesis block generation without channels
2018-03-04 14:32:30 -07:00
b1e9512f44
Rename end_hash to id
2018-03-04 07:50:26 -07:00
608def9c78
Consolidate imports
2018-03-04 07:28:51 -07:00
9b26892bae
Add a demo app to generate the genesis file
2018-03-04 01:21:40 -07:00
572475ce14
Load the genesis block
2018-03-04 00:15:17 -07:00
b8655e30d4
Make client-demo standalone
...
And remove deposit() methods from the API. Those should only be
used on the server to bootstrap.
2018-03-03 21:15:51 -07:00
7cf0d55546
Remove optional 'from' field
2018-03-03 20:41:07 -07:00
cebcb5b92d
Start genesis with a Tick, so that its hash can be used to bootstrap verification
2018-03-03 19:57:22 -07:00
11a0f96f5e
Add command-line tool for generating a genesis block
2018-03-03 17:35:05 -07:00
1c9b5ab53c
Report performance of signature verification too
2018-03-03 11:59:34 -07:00
c10faae3b5
More readable metrics
2018-03-03 11:52:50 -07:00
2104dd5a0a
Fix benchmark
...
Was measuring the creation of the iterator, not running it.
2018-03-03 11:45:23 -07:00
d8c50b150c
Move key generation and signing from transaction benchmark
...
Key generation, signing and verification are not the performance
bottleneck. Something is probably wrong here.
2018-03-03 11:11:46 -07:00
a148454376
Update readme
2018-03-02 12:07:05 -07:00
0d0fee1ca1
Sign Claim's 'to' field
...
Otherwise, the accountant will treat deposits of the same amount as
duplicates.
2018-03-02 11:46:22 -07:00
a0410c4677
Pipe all Claim constructors through a function
2018-03-02 10:58:43 -07:00
8fe464cfa3
Rename Claim's key field to match same field in Transaction
2018-03-02 10:47:21 -07:00
32d677787b
Reduce transactions sent by demo
...
We don't do retries yet, so keep tx count to something that won't
trigger any packet loss.
2018-03-02 10:35:38 -07:00
1b77e8a69a
Move Event into its own crate
...
The log crate was starting to be the catch-all for all things
related to entries, events, signatures, and hashes. This split
shows us that:
* Event depends only on signatures, not on hashes [directly]
* All event testing was done via log testing (shame on me)
* Accounting depends only on events
2018-03-02 08:43:57 -07:00
9e16937914
Delete the Discovery event
...
Not useful to the accountant.
2018-03-01 17:02:41 -07:00
570e71f050
Check for duplicate signatures
...
TODO: have client add recent hash to each message
2018-03-01 14:07:39 -07:00
c9cc4b4369
Switch to UDP from TCP
...
And remove all the sleep()'ing around.
2018-03-01 13:47:53 -07:00
4610de8fdd
Switch to sync_channel to preserve order
2018-02-28 19:33:28 -07:00
3fcc2dd944
Add testnode
...
Fixes #20
2018-02-28 18:05:20 -07:00
8299bae2d4
Add accountant stub
2018-02-28 16:01:12 -07:00
604ccf7552
Add network interface for accountant
2018-02-28 14:00:04 -07:00
c6134cc25b
Allow the historian to track ownership of any type of data
2018-02-26 17:01:22 -07:00
8b0b8efbcb
Allow Entry to hold events of any kind of data
2018-02-26 15:37:33 -07:00
97449cee43
Allow events to hold any kind of data
2018-02-26 15:31:01 -07:00
b8d52cc3e4
Make the Discovery event into a struct instead of a tuple
2018-02-24 11:15:03 -07:00
29a607427d
Rename UserDataKey to Discovery
...
From the perspective of the log, when some data's hash is added,
that data is "discovered" by the historian. Another event
might be a "claim" that some signed data belongs to the owner of a
public key.
2018-02-24 05:25:19 -07:00
7871b38c80
Update demo to use self-ticking logger
2018-02-21 11:52:03 -07:00
b34d2d7dee
Allow the logger to inject Tick events on its own
2018-02-21 11:33:42 -07:00
d97112d7f0
Explain proof-of-history in the readme
...
Also:
* Hash userdata so that verification works as the readme describes.
* Drop itertools package. Found a way to use std::iter instead.
Fixes #8
2018-02-20 14:04:49 -07:00
0297edaf1f
Use sha256 hashes instead of Rust's builtin hasher.
...
Causes a 20x performance degradation. Enabling asm did not
speed things up.
2018-02-19 16:23:53 -07:00
10a70a238b
Cleanup demo
2018-02-19 12:25:57 -07:00
d88d1b2a09
Reset historian's hasher between events
...
Hasher will generate different hashes for the same input if it
had already generated a hash.
Also add a binary to ensure the example in the README works.
2018-02-19 12:03:06 -07:00
Greg Fitzgerald
Code Cobain
Robert Kelly
Tyera Eulberg
Stephen Akridge
Anatoly Yakovenko
rleungx