* updates some rate limits
* WOAH FRIDAY FAILURES
* remove secret code, DONT LOOK
* types.Number, rv ob num, fr changes
* ever so slight neaten
* Remove FundingRateInfo for CMF only
* revert to using the funding rate intervals
* don't need to verify if checksum available; don't retroactively verify snapshots
* glorious: nits + allow checksum/validation to be done on updateByIDAndAction
* possible slice manipulation issue fix for test
* thrasher: nit
* this should fix it now
* for subsystem usage disallow deploying a book twice
* reverts strict policy using deploy, it's used in a bunch of tests I don't want to touch just yet, left a note
* rm unused variable
* glorious: nits
---------
Co-authored-by: Ryan O'Hara-Reid <ryan.oharareid@thrasher.io>
* bybit: Add time field for orderbook and handle correct timings
* use UpdateID as this is the localised ID for this pair, rm check result.UpdateID == 1 just use snapshot
* general quick fixes across from #1672 which can be used straight away
* fix test
* glorious: nits
* thrasher: nits
* thrasher: nits thanks @thrasher-
---------
Co-authored-by: Ryan O'Hara-Reid <ryan.oharareid@thrasher.io>
* Bybit: Subscription configuration for spot
* Bybit: Enable candles ws sub by default
* Orderbook: Use a RW mutex for depth
* Orderbook: Fix race on depth.VerifyOrderbook
Despite being protected by an ob level mutex, this needed to privatise
and protect the option var.
The subscription pairs do not need to be validated as enabled or
available. The check was just belt-and-braces and didn't have a specific
use-case in mind.
Coinbase has a use-case for wanting to subscribe to BTC-USD when it's
not enabled.
Moreover, it shouldn't be our job to check this. You want a sub expanded
with these pairs? Fine. Done. If it doesn't work, you can work out why
* Exchanges: Allow empty batches in ParallelChanOp
In keeping with both common.Batch and "It's not my responsibility",
ParallelChanOp should just do nothing when given an empty list (and
implicitly an empty batch size.
Whatever it's going to do, it'll delegate to common.Batch,
and this allows us to just inline calls:
```
return common.AppendError(
b.ParallelChanOp(subs.Public(), func(l subscription.List) error { return b.manageSubs(wsSubscribeMethod, l, wsPublicStream) }, len(subs)),
b.ParallelChanOp(subs.Private(), func(l subscription.List) error { return b.manageSubs(wsSubscribeMethod, l, wsPrivateStream) }, len(subs)),
)
```
* Bitmex: Test config updates
* Bitmex: Sub Templating
* Bitmex: Enable websocket for tests
* Bitmex: Handle subscription errors
This switches to multiplexing so that we know which errors belong to
which stream, particularly for the auth attempt
* Bitmex: Fix ws order side err going to data stream
Shouldn't fall into classification error if it's actually a parsing
error
* Cherry_pickable
* gateio/websocket: use millisecond time for more accurate push time
---------
Co-authored-by: Ryan O'Hara-Reid <ryan.oharareid@thrasher.io>
* gateio: Add multi asset websocket support WIP.
* meow
* Add tests and shenanigans
* integrate flushing and for enabling/disabling pairs from rpc shenanigans
* some changes
* linter: fixes strikes again.
* Change name ConnectionAssociation -> ConnectionCandidate for better clarity on purpose. Change connections map to point to candidate to track subscriptions for future dynamic connections holder and drop struct ConnectionDetails.
* Add subscription tests (state functional)
* glorious:nits + proxy handling
* Spelling
* linter: fixerino
* instead of nil, dont do nil.
* clean up nils
* cya nils
* don't need to set URL or check if its running
* stop ping handler routine leak
* * Fix bug where reader routine on error that is not a disconnection error but websocket frame error or anything really makes the reader routine return and then connection never cycles and the buffer gets filled.
* Handle reconnection via an errors.Is check which is simpler and in that scope allow for quick disconnect reconnect without waiting for connection cycle.
* Dial now uses code from DialContext but just calls context.Background()
* Don't allow reader to return on parse binary response error. Just output error and return a non nil response
* Allow rollback on connect on any error across all connections
* fix shadow jutsu
* glorious/gk: nitters - adds in ws mock server
* linter: fix
* fix deadlock on connection as the previous channel had no reader and would hang connection reader for eternity.
* gk: nits
* Leak issue and edge case
* gk: nits
* gk: drain brain
* glorious: nits
* Update exchanges/stream/websocket.go
Co-authored-by: Scott <gloriousCode@users.noreply.github.com>
* glorious: nits
* add tests
* linter: fix
* After merge
* Add error connection info
* Fix edge case where it does not reconnect made by an already closed connection
* stream coverage
* glorious: nits
* glorious: nits removed asset error handling in stream package
* linter: fix
* rm block
* Add basic readme
* fix asset enabled flush cycle for multi connection
* spella: fix
* linter: fix
* Add glorious suggestions, fix some race thing
* reinstate name before any routine gets spawned
* stop on error in mock tests
* glorious: nits
* glorious: nits found in CI build
* Add test for drain, bumped wait times as there seems to be something happening on macos CI builds, used context.WithTimeout because its instant.
* mutex across shutdown and connect for protection
* lint: fix
* test time withoffset, reinstate stop
* fix whoops
* const trafficCheckInterval; rm testmain
* y
* fix lint
* bump time check window
* stream: fix intermittant test failures while testing routines and remove code that is not needed.
* spells
* cant do what I did
* protect race due to routine.
* update testURL
* use mock websocket connection instead of test URL's
* linter: fix
* remove url because its throwing errors on CI builds
* connections drop all the time, don't need to worry about not being able to echo back ws data as it can be easily reviewed _test file side.
* remove another superfluous url thats not really set up for this
* spawn overwatch routine when there is no errors, inline checker instead of waiting for a time period, add sleep inline with echo handler as this is really quick and wanted to ensure that latency is handing correctly
* linter: fixerino uperino
* glorious: panix
* linter: things
* whoops
* defer lock and use functions that don't require locking in SetProxyAddress
* lint: fix
* thrasher: nits
---------
Co-authored-by: shazbert <ryan.oharareid@thrasher.io>
Co-authored-by: Scott <gloriousCode@users.noreply.github.com>
* Convert: Fix TimeFromUnixTimestampDecimal using local
All parsed times should be in UTC
* Subscriptions: Add IgnoringAssetsKey
* Tests: Pass tb to curried WS handlers
* Websocket: Make ErrNoMessageListener a public error
* Kraken: Fix URLMap ignored for websocket URLs
* Kraken: Move SeedAssets from Setup to Bootstrap
Having SeedAssets in Setup is cruel and unusual because it calls the
API. Most other interactive data seeding happens in Bootstrap.
This made it so that fixing and creating unit tests for Kraken was
painfully slow, particularly on flaky internet.
* Kraken: Remove convert test
Duplicate of convert_test.go TestTimeFromUnixTimestampDecimal
* Kraken: Test config upgrades
* Kraken: Sub Channel improvements
* Use Websocket subscriptionChannels instead of local slice
* Remove ChannelID - Deprecated in docs
* Simplify ping handlers and hardcodes message
* Add Depth as configurable orderbook channel param
* Simplify auth/non-auth channel updates
* Add configurable Book depth
* Add configurable Candle timeframes
Kraken: Simplify all WS handlers with reqId
* Kraken: Subscription templating
* Generate N+ subs for pairs
If we generate one sub for all pairs, but then fan it out in the
responses, we end up with a mis-match between the sub store and
GenerateSubs, and when we do FlushChannels it will try to resub
everything again.
* Kraken: Rename channelName var throughout
Avoid shadowing func of same name
* Kraken: Add TestEnforceStandardChannelNames
* Websocket: Fix Resubscribe erroring Duplicate
* consolidate type to types package and share across code base
* rm convert type and convert codebase
* rm irrelavant test cases
* Fix tests
* glorious nits
* Update exchanges/gateio/gateio_types.go
Co-authored-by: Adrian Gallagher <adrian.gallagher@thrasher.io>
* thrasher: nits
---------
Co-authored-by: Ryan O'Hara-Reid <ryan.oharareid@thrasher.io>
Co-authored-by: Adrian Gallagher <adrian.gallagher@thrasher.io>
