* Add NetworkService::send_notifications
* Doc
* Doc
* API adjustment
* Address concerns
* Make it compile
* Start implementation
* Progress in the implementation
* Change implementation strategy again
* More work before weekend
* Finish changes
* Minor doc fix
* Revert some minor changes
* Apply suggestions from code review
* GroupError -> NotifsHandlerError
* Apply suggestions from code review
Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
* state_transition_waker -> close_waker
* Apply suggestions from code review
Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
* Finish renames in service.rs
* More renames
* More review suggestsions applied
* More review addressing
* Final change
* 512 -> 2048
Co-authored-by: Roman Borschel <romanb@users.noreply.github.com>
* Remove service components and add build_network, build_offchain_workers etc
* Improve transaction pool api
* Remove commented out line
* Add PartialComponents
* Add BuildNetworkParams, documentation
* Remove unused imports in tests
* Apply suggestions from code review
Co-authored-by: Nikolay Volf <nikvolf@gmail.com>
* Remove unused imports in node-bench
Co-authored-by: Nikolay Volf <nikvolf@gmail.com>
* grandpa: remove unused methods to convert digest
* grandpa: add root extrinsic for scheduling forced change
* grandpa: add benchmark for schedule_forced_change
* grandpa: don't take authority weight in schedule_forced_change
* grandpa: add const for default forced change delay
* grandpa: adjust weights after benchmark on ref hardware
* grandpa: fix cleanup of forced changes on standard change application
* grandpa: replace schedule_forced_change with note_stalled
* grandpa: always trigger a session change when the set is stalled
* grandpa: fix bug on set id mutation after failed scheduled change
* grandpa: take delay as parameter in note_stalled
* grandpa: fix tests
* grandpa: fix cleanup of forced changes
* grandpa: add test for forced changes cleanup
* grandpa: add test for session rotation set id
* grandpa: add test for scheduling of forced changes on new session
* Ensure we close old file handles hold by sysinfo
* Dropping is needed unfortunately
* enable process refreshing, ignore result from refresh_process
* jumping to proposed patch
* switch to latest sysinfo
* Switch to serde_json::Value for Visitor values
* add tests
* switch Visitor to use Map
* refactor change names
* \n
* update integration test
* use discrete maps for each type of recorded value
* update integration test
* add docs
* Update client/tracing/src/lib.rs
Co-authored-by: Kian Paimani <5588131+kianenigma@users.noreply.github.com>
* add docs
Co-authored-by: Matt Rutherford <mattrutherford@users.noreply.github.com>
Co-authored-by: Kian Paimani <5588131+kianenigma@users.noreply.github.com>
* Initial commit
Forked at: 74655edebc
Parent branch: origin/master
* Move task_manager.rs to mod.rs
* Graceful shutdown for the task manager
* Await all background task JoinHandle at the same time
* Add tests
* Make future() wait also for exit signal + fix essential task failed
Probably related to https://github.com/paritytech/cumulus/issues/111
* add comments for non-obvious code
* Use clean_shutdown() in sc-cli
* Adapt code and upgrade tokio in sc-cli
* cleanup spacing in doc
* Add license
* I guess actually running the clean shutdown would be a good idea
* fix tests
* Update client/cli/src/runner.rs
Co-authored-by: Benjamin Kampmann <ben@gnunicorn.org>
* Improve error logging
* disable other tests (can't reproduce on my machine)
* Revert "disable other tests (can't reproduce on my machine)"
This reverts commit c133c590f33c253123ba0555ce719a71ededd60d.
* It is possible that the tasks are ended first
* Revert "It is possible that the tasks are ended first"
This reverts commit 502aba4a49fb5d892e704c412b8a81768a3f2c71.
* Use single threaded scheduler for more predictability
* enable_time
* Revert "enable_time"
This reverts commit 4e152140764a4bddeedff06a4e36ec701909e8c7.
* Revert "Use single threaded scheduler for more predictability"
This reverts commit ee5e13c5f13ff71e012dcda13579dffeb15f8ffc.
* Revert "Revert "It is possible that the tasks are ended first""
This reverts commit 1b91a8ca3eebbdc18be199c8ca188e88669ae649.
* This cannot be verified either with a threaded pool
* Apply suggestions from code review
Co-authored-by: Bastian Köcher <bkchr@users.noreply.github.com>
Co-authored-by: Benjamin Kampmann <ben@parity.io>
Co-authored-by: Benjamin Kampmann <ben@gnunicorn.org>
Co-authored-by: Bastian Köcher <bkchr@users.noreply.github.com>
* Move the legacy protocol handshake to the legacy substream
* Fix tests
* Remove line that wasn't supposed to be committed
* Remove hack
* Rework how it's done
* Some little changes
* update_chain wasn't doing its thing
* Fix service tests not calling update_chain
* Update client/network/src/protocol/generic_proto/behaviour.rs
Co-authored-by: Max Inden <mail@max-inden.de>
* [WIP]
* Revert "[WIP]"
This reverts commit 2b892e6a7637c0b1297e6ecdbb919321c9098ff5.
* Update client/network/src/protocol.rs
Co-authored-by: Max Inden <mail@max-inden.de>
* Fix received message not being handshake
* Update client/network/src/protocol/generic_proto/behaviour.rs
Co-authored-by: Max Inden <mail@max-inden.de>
Co-authored-by: Max Inden <mail@max-inden.de>
We should check that `serde` is present when using
`sc-chains-spec-derive`, because we require this for the generated
struct. This also switches the generated `derive` instruction to import
`Serialize`/`Deserialize` by path to not require the user to import
these derive macros.
Instead of requiring the `LogRotationOpt`, it now requires an
`Option<LogRotationOpt>`. This makes it much more easy to use the
interface when someone isn't interested on the `LogRotationOpt`'s
* Simplify a few chain components creation APIs related to the service
* Fix basic-authorship doc tests
* Remove DefaultQueue
* Update client/service/src/builder.rs
Co-authored-by: André Silva <123550+andresilva@users.noreply.github.com>
* Move ExecutionExtensions comment around
* Remove unused BlakeTwo256
Co-authored-by: André Silva <123550+andresilva@users.noreply.github.com>
* Improve transaction submission
Before this pr the transaction pool validated each transaction, even if
the transaction was already known to the pool. This pr changes the
behavior to first check if we are already aware of a transaction and
thus, to only validate them if we don't know them yet. However, there is
still the possibility that a given transaction is validated multiple
times. This can happen if the transaction is added the first time, but
is not yet validated and added to the validated pool.
Besides that, this pr fixes the wrong metrics of gossiped transactions
in the network. It also moves some metrics to the transaction pool api,
to better track when a transaction actually is scheduled for validation.
* Make sure we don't submit the same transaction twice from the network concurrently
* Remove added listener call
* Feedback
* Ignore banned on resubmit
Information retrieved via `DebugInfoBehaviour` is not only used for
debugging purposes, e.g. disconnecting from nodes not responding to
pings, using external addresses retrieved via indentify, ...
In order for the name to reflect the usage of the module this commit
renames it.
* Upgrade `kvdb-*`, `trie-db` and `memory-db`
The updates of `trie-db` and `memory-db` are important, as they fix the
non-deterministic build of Polkadot/Substrate.
* Change `trie-db` version
* Update test-utils/runtime/Cargo.toml
Co-authored-by: Andronik Ordian <write@reusable.software>
* Update primitives/trie/Cargo.toml
Co-authored-by: Andronik Ordian <write@reusable.software>
* Update `Cargo.lock` and `trie-bench`
* Fix UI tests
* Switch to fixed version of memory-db
Co-authored-by: Andronik Ordian <write@reusable.software>
* slots: create primitives crate for consensus slots
* offences: add method to check if an offence is unknown
* babe: initial equivocation reporting implementation
* babe: organize imports
* babe: working equivocation reporting
* babe: add slot number to equivocation proof
* session: move duplicate traits to session primitives
* babe: move equivocation stuff to its own file
* offences: fix test
* session: don't have primitives depend on frame_support
* babe: use opaque type for key owner proof
* babe: cleanup client equivocation reporting
* babe: cleanup equivocation code in pallet
* babe: allow sending signed equivocation reports
* node: fix compilation
* fix test compilation
* babe: return bool on check_equivocation_proof
* babe: add test for equivocation reporting
* babe: add more tests
* babe: add test for validate unsigned
* babe: take slot number in generate_key_ownership_proof API
* babe: add benchmark for equivocation proof checking
* session: add benchmark for membership proof checking
* offences: fix babe benchmark
* babe: add weights based on benchmark results
* babe: adjust weights after benchmarking on reference hardware
* babe: reorder checks in check_and_report_equivocation
* Move `create_inherents` into the block-builder
This moves the `create_inherents` call into the block-builder. This has
the advantage that `create_inherents` will be able to reuse the same
context that will be used when applying the extrinsics and we also save
one call to `on_initialize`. To make sure that `create_inherents` does
not modify any state, we execute it in a transaction that is
rolled-back after doing the runtime call.
* Feedback and build fix
* Update primitives/runtime/src/lib.rs
Co-authored-by: Sergei Shulepov <sergei@parity.io>
* Update client/block-builder/src/lib.rs
Co-authored-by: Sergei Shulepov <sergei@parity.io>
* Update SubstrateCli to return String
* Add default implementation for executable_name()
* Use display instead of PathBuf
* Get file_name in default impl of executable_name
* Remove String::from and use .into()
* Use default impl for executable_name()
* Use .as_str() and remove useless .to_string()
* Update only sp-io when running companion build
* Remove unneeded update of sp-io in CI
Co-authored-by: Cecile Tonglet <cecile@parity.io>
* Initial commit
Forked at: ece0364170
Parent branch: origin/master
* Drop the tokio runtime before the task_manager
The tokio runtime must be dropped before the task_manager. Otherwise the objects the task_manager
keep alive are dropped before the tasks are finished.
Gavin Wood