2026-05-19 11:54:54 -07:00
|
|
|
/// GroupV1 is a conversationType which provides effecient handling of multiple participants
|
|
|
|
|
/// Properties:
|
|
|
|
|
/// - Harvest Now Decrypt Later (HNDL) protection provided by XWING
|
|
|
|
|
/// - Multiple
|
|
|
|
|
use blake2::{Blake2b, Digest, digest::consts::U6};
|
|
|
|
|
use chat_proto::logoschat::encryption::{EncryptedPayload, Plaintext, encrypted_payload};
|
2026-05-25 11:54:52 +08:00
|
|
|
use chat_proto::logoschat::reliability::ReliablePayload;
|
2026-05-19 11:54:54 -07:00
|
|
|
use openmls::prelude::tls_codec::Deserialize;
|
|
|
|
|
use openmls::prelude::*;
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
use prost::Message as _;
|
2026-06-10 06:59:04 -07:00
|
|
|
use shared_traits::IdentIdRef;
|
2026-05-19 11:54:54 -07:00
|
|
|
|
2026-06-11 21:07:11 +08:00
|
|
|
use crate::account_directory::{AccountDirectory, resolve_device_ids};
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
use crate::inbox_v2::MlsProvider;
|
|
|
|
|
use crate::service_context::{ExternalServices, ServiceContext};
|
2026-06-10 06:59:04 -07:00
|
|
|
|
2026-05-19 11:54:54 -07:00
|
|
|
use crate::{
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
DeliveryService, IdentityProvider,
|
|
|
|
|
conversation::{ChatError, Convo, GroupConvo},
|
feat: introduce client event system (#106)
* chore(flake): accept extra system attr; add perl for openssl-sys build
forAllSystems calls the lambda with {system, pkgs}; strict
destructuring requires `..` to ignore the system attribute.
`pkgs.perl` is needed because openssl-sys is pulled vendored via
libsqlite3-sys / rusqlite / chat-sqlite, and its `perl Configure`
step needs FindBin.pm, which Fedora's system perl doesn't ship.
* feat: introduce client event system
- Core processing yields a `PayloadOutcome` enum — `Empty`, `Convo`, or
`Inbox`. `ConvoOutcome` carries a conversation id and an optional
decrypted `Content`; `InboxOutcome` adds a `NewConversation`
(id + `ConversationClass`) for a peer-initiated conversation.
- Client translates `PayloadOutcome` into app-facing `Vec<Event>`
(`ConversationStarted`, `MessageReceived`) at the boundary, so the
application loop sees discrete events rather than core types.
- MLS group welcomes produce a `ConversationStarted` event with no
initial content, fixing the silent-group-join case where the inbox
layer dropped the observation.
- C FFI exposes an `EventList` opaque type with indexed accessors and
an `Invalid` sentinel for out-of-bounds / non-applicable reads.
- Symmetric `Inbox` / `InboxV2` handlers: both return
`Result<InboxOutcome, _>` and own the persistence + ephemeral-key
cleanup for the conversations they create.
- Updated and simplified `docs/adr/0001-client-event-system.md`.
* chore(flake): bump nixpkgs to nixos-unstable-small
Temporary. The two crates.io UA fixes (NixOS/nixpkgs#512735 for
fetchCargoVendor's python-requests UA, NixOS/nixpkgs#524985 for
importCargoLock's curl UA) haven't propagated to nixos-unstable yet.
Switch to nixos-unstable-small and force logos-delivery to follow so
the smoketest gets the same fix. Revert once nixos-unstable catches up.
Refs:
- https://github.com/rust-lang/crates.io/issues/13482
- https://github.com/rust-lang/crates.io/issues/13783
- https://crates.io/data-access
2026-05-28 23:51:15 +02:00
|
|
|
outcomes::{Content, ConvoOutcome},
|
2026-05-19 11:54:54 -07:00
|
|
|
service_traits::KeyPackageProvider,
|
feat: introduce client event system (#106)
* chore(flake): accept extra system attr; add perl for openssl-sys build
forAllSystems calls the lambda with {system, pkgs}; strict
destructuring requires `..` to ignore the system attribute.
`pkgs.perl` is needed because openssl-sys is pulled vendored via
libsqlite3-sys / rusqlite / chat-sqlite, and its `perl Configure`
step needs FindBin.pm, which Fedora's system perl doesn't ship.
* feat: introduce client event system
- Core processing yields a `PayloadOutcome` enum — `Empty`, `Convo`, or
`Inbox`. `ConvoOutcome` carries a conversation id and an optional
decrypted `Content`; `InboxOutcome` adds a `NewConversation`
(id + `ConversationClass`) for a peer-initiated conversation.
- Client translates `PayloadOutcome` into app-facing `Vec<Event>`
(`ConversationStarted`, `MessageReceived`) at the boundary, so the
application loop sees discrete events rather than core types.
- MLS group welcomes produce a `ConversationStarted` event with no
initial content, fixing the silent-group-join case where the inbox
layer dropped the observation.
- C FFI exposes an `EventList` opaque type with indexed accessors and
an `Invalid` sentinel for out-of-bounds / non-applicable reads.
- Symmetric `Inbox` / `InboxV2` handlers: both return
`Result<InboxOutcome, _>` and own the persistence + ephemeral-key
cleanup for the conversations they create.
- Updated and simplified `docs/adr/0001-client-event-system.md`.
* chore(flake): bump nixpkgs to nixos-unstable-small
Temporary. The two crates.io UA fixes (NixOS/nixpkgs#512735 for
fetchCargoVendor's python-requests UA, NixOS/nixpkgs#524985 for
importCargoLock's curl UA) haven't propagated to nixos-unstable yet.
Switch to nixos-unstable-small and force logos-delivery to follow so
the smoketest gets the same fix. Revert once nixos-unstable catches up.
Refs:
- https://github.com/rust-lang/crates.io/issues/13482
- https://github.com/rust-lang/crates.io/issues/13783
- https://crates.io/data-access
2026-05-28 23:51:15 +02:00
|
|
|
types::AddressedEncryptedPayload,
|
2026-05-19 11:54:54 -07:00
|
|
|
};
|
|
|
|
|
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
pub struct GroupV1Convo {
|
2026-05-19 11:54:54 -07:00
|
|
|
mls_group: MlsGroup,
|
|
|
|
|
convo_id: String,
|
|
|
|
|
}
|
|
|
|
|
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
impl std::fmt::Debug for GroupV1Convo {
|
2026-05-19 11:54:54 -07:00
|
|
|
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
|
|
|
|
|
f.debug_struct("GroupV1Convo")
|
|
|
|
|
.field("convo_id", &self.convo_id)
|
|
|
|
|
.field("mls_epoch", &self.mls_group.epoch())
|
|
|
|
|
.finish_non_exhaustive()
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
impl GroupV1Convo {
|
2026-05-19 11:54:54 -07:00
|
|
|
// Create a new conversation with the creator as the only participant.
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
pub fn new<S: ExternalServices>(cx: &mut ServiceContext<S>) -> Result<Self, ChatError> {
|
2026-06-15 13:15:18 -07:00
|
|
|
let config = Self::mls_create_config(cx);
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
let mls_group = MlsGroup::new(
|
|
|
|
|
&cx.mls_provider,
|
|
|
|
|
&cx.mls_identity,
|
|
|
|
|
&config,
|
|
|
|
|
cx.mls_identity.get_credential(),
|
|
|
|
|
)
|
|
|
|
|
.unwrap();
|
2026-05-19 11:54:54 -07:00
|
|
|
let convo_id = hex::encode(mls_group.group_id().as_slice());
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
Self::subscribe(&mut cx.ds, &convo_id)?;
|
2026-05-19 11:54:54 -07:00
|
|
|
|
|
|
|
|
Ok(Self {
|
|
|
|
|
mls_group,
|
|
|
|
|
convo_id,
|
|
|
|
|
})
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Constructs a new conversation upon receiving a MlsWelcome message.
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
pub fn new_from_welcome<S: ExternalServices>(
|
|
|
|
|
cx: &mut ServiceContext<S>,
|
2026-05-19 11:54:54 -07:00
|
|
|
welcome: Welcome,
|
|
|
|
|
) -> Result<Self, ChatError> {
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
let mls_group =
|
|
|
|
|
StagedWelcome::build_from_welcome(&cx.mls_provider, &Self::mls_join_config(), welcome)
|
2026-05-19 11:54:54 -07:00
|
|
|
.unwrap()
|
|
|
|
|
.build()
|
|
|
|
|
.unwrap()
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
.into_group(&cx.mls_provider)
|
|
|
|
|
.unwrap();
|
2026-05-19 11:54:54 -07:00
|
|
|
|
|
|
|
|
let convo_id = hex::encode(mls_group.group_id().as_slice());
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
Self::subscribe(&mut cx.ds, &convo_id)?;
|
2026-05-19 11:54:54 -07:00
|
|
|
|
|
|
|
|
Ok(Self {
|
|
|
|
|
mls_group,
|
|
|
|
|
convo_id,
|
|
|
|
|
})
|
|
|
|
|
}
|
|
|
|
|
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
pub fn load<S: ExternalServices>(
|
|
|
|
|
cx: &mut ServiceContext<S>,
|
2026-05-19 11:54:54 -07:00
|
|
|
convo_id: String,
|
|
|
|
|
group_id: GroupId,
|
|
|
|
|
) -> Result<Self, ChatError> {
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
let mls_group = MlsGroup::load(cx.mls_provider.storage(), &group_id)
|
2026-05-19 11:54:54 -07:00
|
|
|
.map_err(ChatError::generic)?
|
|
|
|
|
.ok_or_else(|| ChatError::NoConvo("mls group not found".into()))?;
|
|
|
|
|
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
Self::subscribe(&mut cx.ds, &convo_id)?;
|
2026-05-19 11:54:54 -07:00
|
|
|
|
|
|
|
|
Ok(GroupV1Convo {
|
|
|
|
|
mls_group,
|
|
|
|
|
convo_id,
|
|
|
|
|
})
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Configure the delivery service to listen for the required delivery addresses.
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
fn subscribe(ds: &mut impl DeliveryService, convo_id: &str) -> Result<(), ChatError> {
|
2026-05-19 11:54:54 -07:00
|
|
|
ds.subscribe(&Self::delivery_address_from_id(convo_id))
|
|
|
|
|
.map_err(ChatError::generic)?;
|
|
|
|
|
ds.subscribe(&Self::ctrl_delivery_address_from_id(convo_id))
|
|
|
|
|
.map_err(ChatError::generic)?;
|
|
|
|
|
|
|
|
|
|
Ok(())
|
|
|
|
|
}
|
|
|
|
|
|
2026-06-15 13:15:18 -07:00
|
|
|
fn mls_create_config<S: ExternalServices>(cx: &mut ServiceContext<S>) -> MlsGroupCreateConfig {
|
2026-05-19 11:54:54 -07:00
|
|
|
MlsGroupCreateConfig::builder()
|
2026-06-15 13:15:18 -07:00
|
|
|
.ciphersuite(cx.mls_provider.crypto().supported_ciphersuites()[0])
|
2026-05-19 11:54:54 -07:00
|
|
|
.use_ratchet_tree_extension(true) // This is handy for now, until there is central store for this data
|
|
|
|
|
.build()
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
fn mls_join_config() -> MlsGroupJoinConfig {
|
|
|
|
|
MlsGroupJoinConfig::builder().build()
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
fn delivery_address_from_id(convo_id: &str) -> String {
|
|
|
|
|
let hash = Blake2b::<U6>::new()
|
|
|
|
|
.chain_update("delivery_addr|")
|
|
|
|
|
.chain_update(convo_id)
|
|
|
|
|
.finalize();
|
|
|
|
|
hex::encode(hash)
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
fn delivery_address(&self) -> String {
|
|
|
|
|
Self::delivery_address_from_id(&self.convo_id)
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
fn ctrl_delivery_address_from_id(convo_id: &str) -> String {
|
|
|
|
|
let hash = Blake2b::<U6>::new()
|
|
|
|
|
.chain_update("ctrl_delivery_addr|")
|
|
|
|
|
.chain_update(convo_id)
|
|
|
|
|
.finalize();
|
|
|
|
|
hex::encode(hash)
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
fn ctrl_delivery_address(&self) -> String {
|
|
|
|
|
Self::ctrl_delivery_address_from_id(&self.convo_id)
|
|
|
|
|
}
|
|
|
|
|
|
2026-06-11 21:07:11 +08:00
|
|
|
/// Resolve an account to a KeyPackage for *every* device it authorizes.
|
|
|
|
|
///
|
|
|
|
|
/// First resolves the account to its device ids through the account
|
|
|
|
|
/// directory ([`resolve_device_ids`]), then fetches each device's
|
|
|
|
|
/// KeyPackage. When the account never published a bundle, resolution falls
|
|
|
|
|
/// back to a single device id equal to the account id — the pre-directory
|
|
|
|
|
/// behaviour — so single-device accounts are unaffected.
|
|
|
|
|
fn key_packages_for_account(
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
&self,
|
2026-06-10 06:59:04 -07:00
|
|
|
ident: IdentIdRef,
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
provider: &impl MlsProvider,
|
2026-06-11 21:07:11 +08:00
|
|
|
registry: &(impl KeyPackageProvider + AccountDirectory),
|
|
|
|
|
) -> Result<Vec<KeyPackage>, ChatError> {
|
|
|
|
|
let device_ids =
|
|
|
|
|
resolve_device_ids(registry, ident).map_err(|e| ChatError::Generic(e.to_string()))?;
|
|
|
|
|
|
|
|
|
|
let mut keypackages = Vec::with_capacity(device_ids.len());
|
|
|
|
|
for device_id in &device_ids {
|
|
|
|
|
let retrieved = registry
|
|
|
|
|
.retrieve(device_id)
|
|
|
|
|
.map_err(|e| ChatError::Generic(e.to_string()))?;
|
|
|
|
|
let Some(keypkg_bytes) = retrieved else {
|
|
|
|
|
return Err(ChatError::Protocol(format!(
|
|
|
|
|
"no keypackage for device {device_id} of account {ident}"
|
|
|
|
|
)));
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
let key_package_in = KeyPackageIn::tls_deserialize(&mut keypkg_bytes.as_slice())?;
|
|
|
|
|
let keypkg = key_package_in.validate(provider.crypto(), ProtocolVersion::Mls10)?; //TODO: P3 - Hardcoded Protocol Version
|
|
|
|
|
keypackages.push(keypkg);
|
|
|
|
|
}
|
|
|
|
|
Ok(keypackages)
|
2026-05-19 11:54:54 -07:00
|
|
|
}
|
|
|
|
|
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
pub fn id(&self) -> &str {
|
2026-05-19 11:54:54 -07:00
|
|
|
&self.convo_id
|
|
|
|
|
}
|
|
|
|
|
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
fn send_message<S: ExternalServices>(
|
2026-05-19 11:54:54 -07:00
|
|
|
&mut self,
|
|
|
|
|
content: &[u8],
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
cx: &ServiceContext<S>,
|
2026-05-19 11:54:54 -07:00
|
|
|
) -> Result<Vec<AddressedEncryptedPayload>, ChatError> {
|
2026-06-10 06:59:04 -07:00
|
|
|
let sender_id = cx.mls_identity.id().as_str();
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
let reliable = cx.causal.on_send(&self.convo_id, sender_id, content);
|
2026-05-25 11:54:52 +08:00
|
|
|
let wire = reliable.encode_to_vec();
|
|
|
|
|
|
2026-05-19 11:54:54 -07:00
|
|
|
let mls_message_out = self
|
|
|
|
|
.mls_group
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
.create_message(&cx.mls_provider, &cx.mls_identity, &wire)
|
2026-05-19 11:54:54 -07:00
|
|
|
.unwrap();
|
|
|
|
|
|
|
|
|
|
let a = AddressedEncryptedPayload {
|
|
|
|
|
delivery_address: self.delivery_address(),
|
|
|
|
|
data: EncryptedPayload {
|
|
|
|
|
encryption: Some(encrypted_payload::Encryption::Plaintext(Plaintext {
|
|
|
|
|
payload: mls_message_out.to_bytes().unwrap().into(),
|
|
|
|
|
})),
|
|
|
|
|
},
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
Ok(vec![a])
|
|
|
|
|
}
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
impl<S: ExternalServices> Convo<S> for GroupV1Convo {
|
|
|
|
|
fn send_content(
|
|
|
|
|
&mut self,
|
|
|
|
|
cx: &mut ServiceContext<S>,
|
|
|
|
|
content: &[u8],
|
|
|
|
|
) -> Result<(), ChatError> {
|
|
|
|
|
let payloads = self.send_message(content, cx)?;
|
|
|
|
|
for payload in payloads {
|
|
|
|
|
cx.ds
|
|
|
|
|
.publish(payload.into_envelope(self.id().into()))
|
|
|
|
|
.map_err(|e| ChatError::Delivery(e.to_string()))?;
|
|
|
|
|
}
|
|
|
|
|
Ok(())
|
|
|
|
|
}
|
2026-05-19 11:54:54 -07:00
|
|
|
|
|
|
|
|
fn handle_frame(
|
|
|
|
|
&mut self,
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
cx: &mut ServiceContext<S>,
|
2026-05-19 11:54:54 -07:00
|
|
|
encoded_payload: EncryptedPayload,
|
feat: introduce client event system (#106)
* chore(flake): accept extra system attr; add perl for openssl-sys build
forAllSystems calls the lambda with {system, pkgs}; strict
destructuring requires `..` to ignore the system attribute.
`pkgs.perl` is needed because openssl-sys is pulled vendored via
libsqlite3-sys / rusqlite / chat-sqlite, and its `perl Configure`
step needs FindBin.pm, which Fedora's system perl doesn't ship.
* feat: introduce client event system
- Core processing yields a `PayloadOutcome` enum — `Empty`, `Convo`, or
`Inbox`. `ConvoOutcome` carries a conversation id and an optional
decrypted `Content`; `InboxOutcome` adds a `NewConversation`
(id + `ConversationClass`) for a peer-initiated conversation.
- Client translates `PayloadOutcome` into app-facing `Vec<Event>`
(`ConversationStarted`, `MessageReceived`) at the boundary, so the
application loop sees discrete events rather than core types.
- MLS group welcomes produce a `ConversationStarted` event with no
initial content, fixing the silent-group-join case where the inbox
layer dropped the observation.
- C FFI exposes an `EventList` opaque type with indexed accessors and
an `Invalid` sentinel for out-of-bounds / non-applicable reads.
- Symmetric `Inbox` / `InboxV2` handlers: both return
`Result<InboxOutcome, _>` and own the persistence + ephemeral-key
cleanup for the conversations they create.
- Updated and simplified `docs/adr/0001-client-event-system.md`.
* chore(flake): bump nixpkgs to nixos-unstable-small
Temporary. The two crates.io UA fixes (NixOS/nixpkgs#512735 for
fetchCargoVendor's python-requests UA, NixOS/nixpkgs#524985 for
importCargoLock's curl UA) haven't propagated to nixos-unstable yet.
Switch to nixos-unstable-small and force logos-delivery to follow so
the smoketest gets the same fix. Revert once nixos-unstable catches up.
Refs:
- https://github.com/rust-lang/crates.io/issues/13482
- https://github.com/rust-lang/crates.io/issues/13783
- https://crates.io/data-access
2026-05-28 23:51:15 +02:00
|
|
|
) -> Result<ConvoOutcome, ChatError> {
|
2026-05-19 11:54:54 -07:00
|
|
|
let bytes = match encoded_payload.encryption {
|
|
|
|
|
Some(encrypted_payload::Encryption::Plaintext(pt)) => pt.payload,
|
|
|
|
|
_ => {
|
|
|
|
|
return Err(ChatError::ProtocolExpectation(
|
|
|
|
|
"None",
|
|
|
|
|
"Some(Encryption::Plaintext)".into(),
|
|
|
|
|
));
|
|
|
|
|
}
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
let mls_message =
|
|
|
|
|
MlsMessageIn::tls_deserialize_exact_bytes(&bytes).map_err(ChatError::generic)?;
|
|
|
|
|
|
|
|
|
|
let protocol_message: ProtocolMessage = mls_message
|
|
|
|
|
.try_into_protocol_message()
|
|
|
|
|
.map_err(ChatError::generic)?;
|
|
|
|
|
|
|
|
|
|
if protocol_message.epoch() < self.mls_group.epoch() {
|
|
|
|
|
// TODO: (P1) Add logging for messages arriving from past epoch.
|
feat: introduce client event system (#106)
* chore(flake): accept extra system attr; add perl for openssl-sys build
forAllSystems calls the lambda with {system, pkgs}; strict
destructuring requires `..` to ignore the system attribute.
`pkgs.perl` is needed because openssl-sys is pulled vendored via
libsqlite3-sys / rusqlite / chat-sqlite, and its `perl Configure`
step needs FindBin.pm, which Fedora's system perl doesn't ship.
* feat: introduce client event system
- Core processing yields a `PayloadOutcome` enum — `Empty`, `Convo`, or
`Inbox`. `ConvoOutcome` carries a conversation id and an optional
decrypted `Content`; `InboxOutcome` adds a `NewConversation`
(id + `ConversationClass`) for a peer-initiated conversation.
- Client translates `PayloadOutcome` into app-facing `Vec<Event>`
(`ConversationStarted`, `MessageReceived`) at the boundary, so the
application loop sees discrete events rather than core types.
- MLS group welcomes produce a `ConversationStarted` event with no
initial content, fixing the silent-group-join case where the inbox
layer dropped the observation.
- C FFI exposes an `EventList` opaque type with indexed accessors and
an `Invalid` sentinel for out-of-bounds / non-applicable reads.
- Symmetric `Inbox` / `InboxV2` handlers: both return
`Result<InboxOutcome, _>` and own the persistence + ephemeral-key
cleanup for the conversations they create.
- Updated and simplified `docs/adr/0001-client-event-system.md`.
* chore(flake): bump nixpkgs to nixos-unstable-small
Temporary. The two crates.io UA fixes (NixOS/nixpkgs#512735 for
fetchCargoVendor's python-requests UA, NixOS/nixpkgs#524985 for
importCargoLock's curl UA) haven't propagated to nixos-unstable yet.
Switch to nixos-unstable-small and force logos-delivery to follow so
the smoketest gets the same fix. Revert once nixos-unstable catches up.
Refs:
- https://github.com/rust-lang/crates.io/issues/13482
- https://github.com/rust-lang/crates.io/issues/13783
- https://crates.io/data-access
2026-05-28 23:51:15 +02:00
|
|
|
return Ok(ConvoOutcome::empty(self.id().to_string()));
|
2026-05-19 11:54:54 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
let processed = self
|
|
|
|
|
.mls_group
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
.process_message(&cx.mls_provider, protocol_message)
|
2026-05-19 11:54:54 -07:00
|
|
|
.map_err(ChatError::generic)?;
|
|
|
|
|
|
feat: introduce client event system (#106)
* chore(flake): accept extra system attr; add perl for openssl-sys build
forAllSystems calls the lambda with {system, pkgs}; strict
destructuring requires `..` to ignore the system attribute.
`pkgs.perl` is needed because openssl-sys is pulled vendored via
libsqlite3-sys / rusqlite / chat-sqlite, and its `perl Configure`
step needs FindBin.pm, which Fedora's system perl doesn't ship.
* feat: introduce client event system
- Core processing yields a `PayloadOutcome` enum — `Empty`, `Convo`, or
`Inbox`. `ConvoOutcome` carries a conversation id and an optional
decrypted `Content`; `InboxOutcome` adds a `NewConversation`
(id + `ConversationClass`) for a peer-initiated conversation.
- Client translates `PayloadOutcome` into app-facing `Vec<Event>`
(`ConversationStarted`, `MessageReceived`) at the boundary, so the
application loop sees discrete events rather than core types.
- MLS group welcomes produce a `ConversationStarted` event with no
initial content, fixing the silent-group-join case where the inbox
layer dropped the observation.
- C FFI exposes an `EventList` opaque type with indexed accessors and
an `Invalid` sentinel for out-of-bounds / non-applicable reads.
- Symmetric `Inbox` / `InboxV2` handlers: both return
`Result<InboxOutcome, _>` and own the persistence + ephemeral-key
cleanup for the conversations they create.
- Updated and simplified `docs/adr/0001-client-event-system.md`.
* chore(flake): bump nixpkgs to nixos-unstable-small
Temporary. The two crates.io UA fixes (NixOS/nixpkgs#512735 for
fetchCargoVendor's python-requests UA, NixOS/nixpkgs#524985 for
importCargoLock's curl UA) haven't propagated to nixos-unstable yet.
Switch to nixos-unstable-small and force logos-delivery to follow so
the smoketest gets the same fix. Revert once nixos-unstable catches up.
Refs:
- https://github.com/rust-lang/crates.io/issues/13482
- https://github.com/rust-lang/crates.io/issues/13783
- https://crates.io/data-access
2026-05-28 23:51:15 +02:00
|
|
|
let content = match processed.into_content() {
|
2026-05-25 11:54:52 +08:00
|
|
|
ProcessedMessageContent::ApplicationMessage(msg) => {
|
|
|
|
|
let reliable = ReliablePayload::decode(msg.into_bytes().as_slice())?;
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
cx.causal.on_receive(&self.convo_id, &reliable);
|
feat: introduce client event system (#106)
* chore(flake): accept extra system attr; add perl for openssl-sys build
forAllSystems calls the lambda with {system, pkgs}; strict
destructuring requires `..` to ignore the system attribute.
`pkgs.perl` is needed because openssl-sys is pulled vendored via
libsqlite3-sys / rusqlite / chat-sqlite, and its `perl Configure`
step needs FindBin.pm, which Fedora's system perl doesn't ship.
* feat: introduce client event system
- Core processing yields a `PayloadOutcome` enum — `Empty`, `Convo`, or
`Inbox`. `ConvoOutcome` carries a conversation id and an optional
decrypted `Content`; `InboxOutcome` adds a `NewConversation`
(id + `ConversationClass`) for a peer-initiated conversation.
- Client translates `PayloadOutcome` into app-facing `Vec<Event>`
(`ConversationStarted`, `MessageReceived`) at the boundary, so the
application loop sees discrete events rather than core types.
- MLS group welcomes produce a `ConversationStarted` event with no
initial content, fixing the silent-group-join case where the inbox
layer dropped the observation.
- C FFI exposes an `EventList` opaque type with indexed accessors and
an `Invalid` sentinel for out-of-bounds / non-applicable reads.
- Symmetric `Inbox` / `InboxV2` handlers: both return
`Result<InboxOutcome, _>` and own the persistence + ephemeral-key
cleanup for the conversations they create.
- Updated and simplified `docs/adr/0001-client-event-system.md`.
* chore(flake): bump nixpkgs to nixos-unstable-small
Temporary. The two crates.io UA fixes (NixOS/nixpkgs#512735 for
fetchCargoVendor's python-requests UA, NixOS/nixpkgs#524985 for
importCargoLock's curl UA) haven't propagated to nixos-unstable yet.
Switch to nixos-unstable-small and force logos-delivery to follow so
the smoketest gets the same fix. Revert once nixos-unstable catches up.
Refs:
- https://github.com/rust-lang/crates.io/issues/13482
- https://github.com/rust-lang/crates.io/issues/13783
- https://crates.io/data-access
2026-05-28 23:51:15 +02:00
|
|
|
Some(Content {
|
|
|
|
|
bytes: reliable.content.to_vec(),
|
|
|
|
|
})
|
2026-05-25 11:54:52 +08:00
|
|
|
}
|
2026-05-19 11:54:54 -07:00
|
|
|
ProcessedMessageContent::StagedCommitMessage(commit) => {
|
|
|
|
|
self.mls_group
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
.merge_staged_commit(&cx.mls_provider, *commit)
|
2026-05-19 11:54:54 -07:00
|
|
|
.map_err(ChatError::generic)?;
|
feat: introduce client event system (#106)
* chore(flake): accept extra system attr; add perl for openssl-sys build
forAllSystems calls the lambda with {system, pkgs}; strict
destructuring requires `..` to ignore the system attribute.
`pkgs.perl` is needed because openssl-sys is pulled vendored via
libsqlite3-sys / rusqlite / chat-sqlite, and its `perl Configure`
step needs FindBin.pm, which Fedora's system perl doesn't ship.
* feat: introduce client event system
- Core processing yields a `PayloadOutcome` enum — `Empty`, `Convo`, or
`Inbox`. `ConvoOutcome` carries a conversation id and an optional
decrypted `Content`; `InboxOutcome` adds a `NewConversation`
(id + `ConversationClass`) for a peer-initiated conversation.
- Client translates `PayloadOutcome` into app-facing `Vec<Event>`
(`ConversationStarted`, `MessageReceived`) at the boundary, so the
application loop sees discrete events rather than core types.
- MLS group welcomes produce a `ConversationStarted` event with no
initial content, fixing the silent-group-join case where the inbox
layer dropped the observation.
- C FFI exposes an `EventList` opaque type with indexed accessors and
an `Invalid` sentinel for out-of-bounds / non-applicable reads.
- Symmetric `Inbox` / `InboxV2` handlers: both return
`Result<InboxOutcome, _>` and own the persistence + ephemeral-key
cleanup for the conversations they create.
- Updated and simplified `docs/adr/0001-client-event-system.md`.
* chore(flake): bump nixpkgs to nixos-unstable-small
Temporary. The two crates.io UA fixes (NixOS/nixpkgs#512735 for
fetchCargoVendor's python-requests UA, NixOS/nixpkgs#524985 for
importCargoLock's curl UA) haven't propagated to nixos-unstable yet.
Switch to nixos-unstable-small and force logos-delivery to follow so
the smoketest gets the same fix. Revert once nixos-unstable catches up.
Refs:
- https://github.com/rust-lang/crates.io/issues/13482
- https://github.com/rust-lang/crates.io/issues/13783
- https://crates.io/data-access
2026-05-28 23:51:15 +02:00
|
|
|
None
|
2026-05-19 11:54:54 -07:00
|
|
|
}
|
|
|
|
|
_ => {
|
|
|
|
|
// TODO: (P2) Log unknown message type
|
feat: introduce client event system (#106)
* chore(flake): accept extra system attr; add perl for openssl-sys build
forAllSystems calls the lambda with {system, pkgs}; strict
destructuring requires `..` to ignore the system attribute.
`pkgs.perl` is needed because openssl-sys is pulled vendored via
libsqlite3-sys / rusqlite / chat-sqlite, and its `perl Configure`
step needs FindBin.pm, which Fedora's system perl doesn't ship.
* feat: introduce client event system
- Core processing yields a `PayloadOutcome` enum — `Empty`, `Convo`, or
`Inbox`. `ConvoOutcome` carries a conversation id and an optional
decrypted `Content`; `InboxOutcome` adds a `NewConversation`
(id + `ConversationClass`) for a peer-initiated conversation.
- Client translates `PayloadOutcome` into app-facing `Vec<Event>`
(`ConversationStarted`, `MessageReceived`) at the boundary, so the
application loop sees discrete events rather than core types.
- MLS group welcomes produce a `ConversationStarted` event with no
initial content, fixing the silent-group-join case where the inbox
layer dropped the observation.
- C FFI exposes an `EventList` opaque type with indexed accessors and
an `Invalid` sentinel for out-of-bounds / non-applicable reads.
- Symmetric `Inbox` / `InboxV2` handlers: both return
`Result<InboxOutcome, _>` and own the persistence + ephemeral-key
cleanup for the conversations they create.
- Updated and simplified `docs/adr/0001-client-event-system.md`.
* chore(flake): bump nixpkgs to nixos-unstable-small
Temporary. The two crates.io UA fixes (NixOS/nixpkgs#512735 for
fetchCargoVendor's python-requests UA, NixOS/nixpkgs#524985 for
importCargoLock's curl UA) haven't propagated to nixos-unstable yet.
Switch to nixos-unstable-small and force logos-delivery to follow so
the smoketest gets the same fix. Revert once nixos-unstable catches up.
Refs:
- https://github.com/rust-lang/crates.io/issues/13482
- https://github.com/rust-lang/crates.io/issues/13783
- https://crates.io/data-access
2026-05-28 23:51:15 +02:00
|
|
|
None
|
2026-05-19 11:54:54 -07:00
|
|
|
}
|
feat: introduce client event system (#106)
* chore(flake): accept extra system attr; add perl for openssl-sys build
forAllSystems calls the lambda with {system, pkgs}; strict
destructuring requires `..` to ignore the system attribute.
`pkgs.perl` is needed because openssl-sys is pulled vendored via
libsqlite3-sys / rusqlite / chat-sqlite, and its `perl Configure`
step needs FindBin.pm, which Fedora's system perl doesn't ship.
* feat: introduce client event system
- Core processing yields a `PayloadOutcome` enum — `Empty`, `Convo`, or
`Inbox`. `ConvoOutcome` carries a conversation id and an optional
decrypted `Content`; `InboxOutcome` adds a `NewConversation`
(id + `ConversationClass`) for a peer-initiated conversation.
- Client translates `PayloadOutcome` into app-facing `Vec<Event>`
(`ConversationStarted`, `MessageReceived`) at the boundary, so the
application loop sees discrete events rather than core types.
- MLS group welcomes produce a `ConversationStarted` event with no
initial content, fixing the silent-group-join case where the inbox
layer dropped the observation.
- C FFI exposes an `EventList` opaque type with indexed accessors and
an `Invalid` sentinel for out-of-bounds / non-applicable reads.
- Symmetric `Inbox` / `InboxV2` handlers: both return
`Result<InboxOutcome, _>` and own the persistence + ephemeral-key
cleanup for the conversations they create.
- Updated and simplified `docs/adr/0001-client-event-system.md`.
* chore(flake): bump nixpkgs to nixos-unstable-small
Temporary. The two crates.io UA fixes (NixOS/nixpkgs#512735 for
fetchCargoVendor's python-requests UA, NixOS/nixpkgs#524985 for
importCargoLock's curl UA) haven't propagated to nixos-unstable yet.
Switch to nixos-unstable-small and force logos-delivery to follow so
the smoketest gets the same fix. Revert once nixos-unstable catches up.
Refs:
- https://github.com/rust-lang/crates.io/issues/13482
- https://github.com/rust-lang/crates.io/issues/13783
- https://crates.io/data-access
2026-05-28 23:51:15 +02:00
|
|
|
};
|
|
|
|
|
Ok(ConvoOutcome {
|
|
|
|
|
convo_id: self.id().to_string(),
|
|
|
|
|
content,
|
|
|
|
|
})
|
2026-05-19 11:54:54 -07:00
|
|
|
}
|
2026-06-15 13:15:18 -07:00
|
|
|
|
|
|
|
|
fn wakeup(&mut self, _: &mut ServiceContext<S>) -> Result<(), ChatError> {
|
|
|
|
|
Ok(())
|
|
|
|
|
}
|
2026-05-19 11:54:54 -07:00
|
|
|
}
|
|
|
|
|
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
impl<S: ExternalServices> GroupConvo<S> for GroupV1Convo {
|
2026-05-19 11:54:54 -07:00
|
|
|
// add_members returns:
|
|
|
|
|
// commit — the Commit message Alice broadcasts to all members
|
|
|
|
|
// welcome — the Welcome message sent privately to each new joiner
|
|
|
|
|
// _group_info — used for external joins; ignore for now
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
fn add_member(
|
|
|
|
|
&mut self,
|
|
|
|
|
cx: &mut ServiceContext<S>,
|
2026-06-10 06:59:04 -07:00
|
|
|
members: &[IdentIdRef],
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
) -> Result<(), ChatError> {
|
2026-05-19 11:54:54 -07:00
|
|
|
if members.len() > 50 {
|
|
|
|
|
// This is a temporary limit that originates from the the De-MLS epoch time.
|
|
|
|
|
return Err(ChatError::Protocol(
|
|
|
|
|
"Cannot add more than 50 Members at a time".into(),
|
|
|
|
|
));
|
|
|
|
|
}
|
|
|
|
|
|
2026-06-11 21:07:11 +08:00
|
|
|
// Resolve each account to a KeyPackage per authorized device and flatten
|
|
|
|
|
// them into one list — every device of every invitee becomes an MLS
|
|
|
|
|
// leaf, so all of a user's installations join the group.
|
|
|
|
|
let mut keypkgs = Vec::with_capacity(members.len());
|
|
|
|
|
for ident in members {
|
|
|
|
|
keypkgs.extend(self.key_packages_for_account(ident, &cx.mls_provider, &cx.registry)?);
|
|
|
|
|
}
|
2026-05-19 11:54:54 -07:00
|
|
|
|
|
|
|
|
let (commit, welcome, _group_info) = self
|
|
|
|
|
.mls_group
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
.add_members(
|
|
|
|
|
&cx.mls_provider,
|
|
|
|
|
&cx.mls_identity,
|
|
|
|
|
keypkgs.iter().as_slice(),
|
|
|
|
|
)
|
2026-05-19 11:54:54 -07:00
|
|
|
.unwrap();
|
|
|
|
|
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
self.mls_group
|
|
|
|
|
.merge_pending_commit(&cx.mls_provider)
|
|
|
|
|
.unwrap();
|
2026-05-19 11:54:54 -07:00
|
|
|
|
|
|
|
|
// TODO: (P3) Evaluate privacy/performance implications of an aggregated Welcome for multiple users
|
|
|
|
|
for account_id in members {
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
cx.mls_provider
|
|
|
|
|
.invite_user(&mut cx.ds, account_id, &welcome)?;
|
2026-05-19 11:54:54 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
let encrypted_payload = EncryptedPayload {
|
|
|
|
|
encryption: Some(encrypted_payload::Encryption::Plaintext(Plaintext {
|
|
|
|
|
payload: commit.to_bytes()?.into(),
|
|
|
|
|
})),
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
let addr_enc_payload = AddressedEncryptedPayload {
|
|
|
|
|
delivery_address: self.ctrl_delivery_address(),
|
|
|
|
|
data: encrypted_payload,
|
|
|
|
|
};
|
|
|
|
|
// Prepare commit message
|
|
|
|
|
// TODO: (P1) Make GroupConvos agnostic to framing so its less error prone and more
|
|
|
|
|
let env = addr_enc_payload.into_envelope(self.convo_id.clone());
|
|
|
|
|
|
refactor(core): replace Rc-based Context with a synchronous, Send-able Core (#123)
Make the conversations core Send so the threaded client can own it behind an
Arc<Mutex<Core>>: a background worker polls the transport and handles inbound
payloads while the application thread issues outbound calls (send, create
conversation). Sharing the core across those two threads means moving it into
the spawned worker, which is only legal if it is Send. Access stays serialized
by the client's Mutex (one thread at a time), so the core needs Send but not
Sync and carries no lock of its own. See
docs/adr/0001-client-event-system.md for the background-poller design.
The Rc<RefCell> service-sharing is what made the core !Send. Context is de-Rc'd
and renamed to Core, owning its services outright and driving the inbox and
conversation primitives with plain &mut self.
- Services (identity, delivery, store, registry, MLS context, causal history)
are bundled into a ServiceContext<S> behind an ExternalServices trait, with
S = (DS, RS, CS). Constructors live on the (DS, RS, CS) form because S cannot
be inferred backwards through S::DS.
- Inbox, InboxV2, PrivateV1Convo, and GroupV1Convo become non-generic and
receive the ServiceContext bundle as a &mut/& parameter; no Rc or
RefCell-as-shared-state remains, so Core is Send whenever its injected
services are.
- Dispatch branches on ConversationKind in one place: Core rebuilds the target
as a Convo<S>/GroupConvo<S> trait object bound to the service bundle, so
conversations never escape the orchestrator.
- CausalHistoryStore drops its Rc, keeping a plain RefCell.
2026-06-08 21:55:33 +02:00
|
|
|
cx.ds
|
2026-05-19 11:54:54 -07:00
|
|
|
.publish(env)
|
|
|
|
|
.map_err(|e| ChatError::Generic(format!("Publish: {e}")))
|
|
|
|
|
}
|
2026-06-15 13:15:18 -07:00
|
|
|
|
|
|
|
|
fn id(&self) -> super::ConversationIdRef<'_> {
|
|
|
|
|
&self.convo_id
|
|
|
|
|
}
|
2026-05-19 11:54:54 -07:00
|
|
|
}
|