It was very confusing that matrix_sdk::store::make_store_config's first
argument was either a path or a database name depending on the build
configuration. ClientBuilder::{sled_store, indexeddb_store} are also
easier to use.
By default, the new `experimental-nodejs` feature is
disabled. However, when enabled, it uses our own fork of
`indexed_db_futures` that make it work on Node.js, and so this entire
`matrix-sdk-indexebdb` crate.
Inspired by the changes in #1013 I was thinking about the use case for `sync_*` and how we handle error cases. Most notably while we give the callback the option to stop the loop, we don't really give an indication to the outside, how to interpret that cancellation: was there a failure? should we restart?
Take e.g. a connectivity issue on the wire, we'd constantly loop and just `warn`, what you might or might not see. Even if you handle that in the `sync_with_result_callback` and thus break the loop, the outer caller now still doesn't know whether everything is honky dory or whether they should restart.
This Changes reworks that area by having all the `sync` return `Result<(), Error>`, where `()` means it was ended by the inner callback (which in `sync()` never occurs) or `Error` is the error either the inner `result_callback` found or the that was coming from the `send` in the first place. Thus allowing us to e.g. back down to sync as it was a dead wire or restart it if there was only a temporary problem. Making all that a just a bit more "rust-y".
… for matrix-sdk, matrix-sdk-base, matrix-sdk-common and matrix-sdk-crypto.
matrix-sdk-indexeddb as well as the JS bindings and wasm_command_bot are
left as-is because they will likely always require JS.
* starting with jack-in
* starting by flying tui
* connecting to real server, showing info
* add .env to gitignore
* infrastructure for tests
* display loading time for syncv2
* minor design updates
* initial sync
* finalise first edition of sliding sync
* directly link to sliding sync and show rooms list
* nicer UI, toggle logs
* passing through sliding sync homeserver
* separate syncs and disable v2 autostart
* selecting rooms
* nicer view
* configurable batches and more default needed events
* selecting rooms
* calculate and show status info per room
* precalculated room stats
* restructure code to allow for cancellation of streams
* finish up merge updates
* fix calculation error in room list len
* cleaning up system flow
* fixing sync up
* new multi-view API
* move sliding sync in separate module
* fixing format
* adding and clearning views
* expose filters and timeline limits
* renamed
* adding room subscriptions to sliding sync
* update summary
* live fetching and subscriptions in jack-in
* subscribe to selected room
* starting to switch to tuireal - using example
* status bar and first linkup
* re-adding rooms
* implementing port for customised update event
* showing details and timeline
* fix formatting
* cleaner UI, updating details quickly
* make it green
* implement other Ops
* proper handling of invalidation
* saving sliding sync results to db, too
* saving new prev_batch field if given
* split events and timeline
* cleaning up
* live updates
* upgrading to latest ruma and matrix-sdk
* Update tui-logger to fix the broken build
* fixing latest ruma sliding-sync-branch updates
* feat: first set of ffi sliding sync
* expose sliding sync via FFI
* implement un-/subscribe
* implement view state updates
* updating to latest JSON format and ruma update
* implementing room selecting for new data model
* fixing room selection
* fixing feature flag requirements for sliding-sync
* fixing style, clippy and docs
* style(sliding-sync): fixing rust format styles
* fix(ffi): fixing sliding sync merge failure
* fix(jack-in): update jack-in to latest ruma
* fix(sliding-sync): need to have a version set before polling proxy
* expose sliding sync builder over ffi
* add SlidingSyncViewBuilder
* add forgotten changes on sdk itself
* new file logger feature on jack for deeper logging
* fix(http-client): log the raw request we are sending out
* feat(sliding-sync): better logging
* fix(sliding-sync): switch to full-listen after reaching live state in full-sync and make sure we replace the correct entries
* feat(ffi): expose sliding sync view ranges
* fix(ffi): fixing sliding sync start_sync loop to actually loop
* feat(sliding-sync): allow lookup of room data
* feat(sliding-sync-ffi): fetching name of room
* feat(ffi): expose unread_notifications of rooms
* feat(ffi): stoppable spawn for sliding sync
* fix(ffi): expose has_unread_notifications on room
* feat(sliding-sync): latest room message
* fix(sliding-sync): update to latest ruma layout
* doc(sliding-sync): adding docs to builder fns
* feat(sliding-sync): extra signal on the view to inform about general updates
* fix(sync-v4-ffi): expose new callbacks via ffi
* fix(sliding-sync): reduce default required states to make things faster
* fix(sliding-sync): fix build params
* feat(jack-in): extended instrumentation
* fix(sliding-sync): unbreak faulty feature-gate
* fix(sliding-sync-ffi): mut where mut is due
* fix(sdk): allow separate homeserver on every request to unbreak using send on client while in sliding sync on a proxy
* fix(jack-in): update to latest dependencies, that work
* feat(ffi): helper to patch sliding sync room into regular room
* style(jack-in): cargo fmt
* fix(sliding-sync): Update to latest ruma changes
* fix(sliding-sync): fix missing FFI updates to latest ruma
* feat(sliding-sync)!: simplify stream cancellation, cancel ffi sync if already existing
* fix: timeline backwards pagination must work without synctoken
* fix(sliding-sync): clarify order of messaes in alive TL; pick correct last item
* fix: update view delegate api for clarity
* style(jack-in): fix cargo warnings
* feat(sliding-sync): update room details
* fix(sliding-sync): only update room info selectively when given
* fix(sliding-sync-ffi): convert and store counts as u32, check against 0 for has notificaitons
* style: cargo fmt, file endings and a few other minor style fixes
* docs(jack-in): improving CLI and Readme
* feat(sliding-sync): allow setting of required event_states on viewbuilder
* style(sliding-sync): docs and minor fixes
* style(sliding-sync): various clippy fixes
* style(jack-in): clippy suggestions applied
* fix(sliding-sync): Delegate becomes observer
* test(sdk): adding test for request config
* docs: Fixing copyright header
* style(ffi): Nicer naming of params for observer
* fix(ffi): sliding sync is not optional for now
* fix(sdk): remove superflusous tracing instrumentation
* fix(sdk): use structured logging
* fix(jack-in): removed unneded log import
* fix(jack-in): use server_name rather than deprecated user_id on ClientBuilder
* style: typo and clippy
* style(sliding-sync): clippy and formatting
* fix(sliding-sync): cleaning up minor syntax issues
* fix: remove unneded feature-definition section
* fix(sliding-sync): minor fixes as per review
* fix(sliding-sync): Make Builders owned
* fix(sliding-sync): more minor style improvements
* fix(sliding-sync): minor style improvements
* fix(sliding-sync): remove homeserver from RequestConfig, use specific internal fn instead
Co-authored-by: Stefan Ceriu <stefanc@matrix.org>
Merge pull request #1013 from matrix-org/ismail/logout
Add logout facilities to `client` and helpers to ffi for tracking the soft-logout issued by a server. Further more this adapts the `sync_with_callback` by adding a new `sync_with_result_callback` that hands the entire `Result` to the callback.
Jonas Platte