New crate air_interpreter_sede introduces a flexible serialization and deserialization infrastructure.
Low level API: the Format trait and related. It includes simple and universal mechanisms for serializing any possible (e.g. implementing the Serde traits) value.
High level API: Representation trait and related. It allows declaring representation types that can serialize and deserialize only certain types (for example, InterpreterDataRepr can serialize only the InterpreterData, but deserialize InterpreterData and Versions` types), producing newtype values as serialization results.
The serialized representation of CallResults now contains byte arrays, not strings, because the new infrastructure serializes to byte arrays only.
Also, the basic support for multicodecs is added.
* Hide `Rc` inside `CID` type, making it cheap to clone.
* Introduce `CidRef` type that abstracts on `CID`'s inner type.
This change makes code cleaner, makes memory more optimal (single allocation vs two allocations) and makes it easier to change CID's internal representation from string to binary.
Implement join behavior for `canon`'s first argument `peer_id`. It
simplifies generating code in Aqua compiler when `peer_id` is defined
by a variable defined in a `par` branch.
Co-authored-by: Mike Voronov <michail.vms@gmail.com>
* feat(execution-engine)!: canon stream map support [fixes VM-301]
Canon stream map is a CRDT-like map structure that contains a
canonicalized stream map and allows indexed access to its
contents. ATM CSM supports both strings and int64 keys.
If `canon` is to be executed on another peer and the particle is to be
sent here, current peer inserts canon request-sent-by state to avoid
repeated requests, similar to `call` instruction.
Refactored stream and stream generation a lot, it introduces the following changes:
- no generation in data anymore, AquaVM relies on generation inside data to stay valid and places value accordingly to it
- stream is internally divided into previous, current, and new values, before, it was one array for all of them
- recursive streams cursors are refactored and rely on new generation values instead
- the Generation enum was refactored and now contains the source of the generation
* feat(aquavm-air)!: `ap` join behavior
Previously, `ap` of an undefined (first) variable was a catchable error.
Now `ap` has join behavior for its first argument.
It should simplify compiling Aqua to AIR.
Closes#632
* chore(testing-framework)!: fix WASM test runner
Native mode was used before because some package used native runner
for its tests.
This PR allows to explicitly select test runner for tests. Many testing-framework
types are now parametrized with a runner type with almost compatible defaults.
* chore(testing-framework): Add `ReleaseWasmAirRunner`
* chore(testing-framework)!: Rename `AirScriptExecutor::simple` to `AirScriptExecutor::from_annotated`.
* chore: release master
* chore: Bump air-interpreter version to 0.40.0
* feat(aquavm-air): Set minimal supported version to 0.40.0
---------
Co-authored-by: Ivan Boldyrev <ivan@fluence.one>
* feat(avm-server)!: keypair and particle ID arguments
Add `&fluence_keypair::KeyPair` argument to `AVM::call` and
`AVMRunner::call`. This value is further forwarded in a deconstructed
form to WASM Air interpreter, but is not used there yet. Also,
`AVMRunner::call` gets `particle_id: String` argument.
feat(air-interpreter)!: `invoke` methods have three new arguments:
`key_format: u8`, `secret_key_bytes: Vec<u8>` and `paritcle_id: String`.
feat(aquavm-air): `air::execute_air` has two three arguments:
`key_format: u8`, `secret_key_bytes: Vec<u8>` and `paritcle_id: String`.
feat(aquavm-air-cli)!: add `--random-key`/`--ed25519-key file` options to AIR CLI.
* feat(avm-server)!: Add `RunnerError::KeypairError`
* chore(bench): Add signature performance benchmarks
These benchmarks contain valid signature, so they should work with
verification out of the box.
---------
Co-authored-by: Artsiom Shamsutdzinau <shamsartem@gmail.com>
Co-authored-by: folex <0xdxdy@gmail.com>
A peer signs the multiset of call results and canon results it has produced.
New field signatures, a map from peer public key to signature, is added to the interpreter data.
Signatures verification is yet to be done.
* `ValueAggregate` refactoring
0. Service results, canon results and literals are constructed as
separate types that are further wrapped with `ValueAggregate`.
1. `ValueAggregate` is enum that contains all the provenance info.
2. Construction methods get provenance information as well.
* Rename CID state field
Prepare to adding a canon CID field: rename `canon_tracker`/`canon_store`
to `canon_element_tracker`/`canon_element_store`.
* Add canon result store/tracker
* Rename some structs that have CIDs inside
Reflect explicitly that they contain CIDs inside:
`CanonResultAggregate` -> `CanonResultCidAggregate`
`ServiceResultAggregate` -> `ServiceResultCidAggregate`
---------
Co-authored-by: Mike Voronov <michail.vms@gmail.com>
* fix(air): Fold-over-scalar values had wrong lambda
Previously, iterator values in fold over scalar inherited tetraplet from
the scalar. But for security guarantees, they should have `.[N]` lambda
added, where N is an element index.
When fold iterates over canon or stream, elements keep their original
tetraplet; tests for that is added.
