Normally the backend is made to have a singleton instance, and handle all profiles.
However, it may be useful to run instances independently (e.g., for development, to run
components in independent processes, for testing, etc).
This was previously possible by using a specific configuration files at backend's root.
This patch introduce a cleaner way by using an isolation mode.
Isolation mode is activated by settings the `LIBERVIA_ISOLATION` environment variable to
the name of the isolated instance. `LIBERVIA_ISOLATION_MODE` can be used to specify one of
the following mode:
- normal (default): config and shared files are in XDG in a `isolated/<name>` subdirectory
(usually `~/.local/share/libervia/isolated/<name>` and
`~/.config/libervia/isolated/<name>`). Runtime files and other instance specific data
(such as D-Bus name) are also isolated. Config uses cascade as usual.
- full: same as above, but config isn't cascading, it's using only its own isolated config
file.
- tmp: same as normal but the local directory is put in a temporary directory, and deleted
on backend startup. It stays on termination, so data can be checked if necessary (this mode is
usually used for development).
This feature will notably be useful to test severless mode on a single machine.
rel 479
Fix embedded bridge and adapt to changes in core code:
- Add `EmbeddedBridgeSettings` so it can be used in settings.
- Add `AIOBridge` and `TwistedBridge` wrappers.
- Fix callbacks, instead of logging them (and breaking the errback chain), they are now
raised using `BridgeException` like in other bridges.
- Ensure that a singleton `Bridge` is used between backend and frontend.
- Init backend on first call to `bridge_connect`.
- Add Python 3.14 fix (as it is normally done in Launcher which is not run with with
`embedded`, we have to do it there).
- Track generated files.
When returning pending queries `Deferred`, a caller could add its own callbacks/errbacks
and modify the result. When another caller was reusing the `Deferred`, it got the modified
result (which could be `None`) instead of the expected `DiscoInfo`.
This has been fixed by creating and new `Deferred` on each request, and chaining it to the
original pending queries using the new `fork_deferred` helper method, so each caller get
its own isolated `Deferred`.
In `fork_deferred`, `cancel` method is used to call the parent Deferred `cancel`, to avoid
a wrong Exception propagation and the chained Deferred to be eventually called twice
(triggering `AlreadyCalledError`).
`cleanup_query` has been moved above `infos_cb` callback, as it should be the first
callback called.
We sometimes to need to call another method in asyncio context when we already are in this
context. With the `@aio` decorator which makes the method return a `Deferred`, it was
necessary to use a ping pong between `Deferred` and `Future`.
This patch adds a `aio_func` attribute which directly returns the wrapped asyncio method,
so it can now be awaited directly.
- correct the case of `dynamicSettingsRegister` and `dynamicSettings` which are `dict`, not
classes.
- Use new `DynamicSettingsCollection` as base class for classes dynamically built during
parsing, to avoid name conflict.
- Use `DynamicSettingsCollection` instead of `dict` in dynamic field (`plugin` and
`component`). This correct inconsistency, and make linters happy.
- Catch `TOMLDecodeError` when parsing config, to show a more readable error message
instead of full backtrace.
New `jid`, `secret` and `autoconnect` field are used for component settings. When used,
backend will automatically create the profile and set its relevant parameters, or check
that existing profile is in sync (and update it if necessary).
This make component installation easier to Libervia, as it is not
necessary anymore to use the CLI frontend to install a component.
rel 476
- D-Bus bridge check if it can import GLib, to raise early a BridgeInitError if it's not
possible
- PB bridge now do lazy import of `reactor` to avoid race condition in Twisted loop
initialisation.
- regenerated bridges.
This patch implements Pubsub Contacts (currently named "End-to-End Encrypted Contacts
Metadata" but it will be most probably renamed soon as following council discussion, e2e
encryption won't be mandatory anymore).
It adds 2 bridge method `ps_contacts_get` and `ps_contact_set`, and handles retrieving
large contacts or groups lists using RSM.
rel 475
Libervia Media is now a Python package instead of files to put in a directory to specify
in settings. The discovery is now automatic, removing the need for settings and enabling
easier installation. Furthermore, it should now work even if the package is zipped.
`libervia_media` can now be specified as a dependency for frontends requiring it.
Singular form is chosen as it makes more sense when configuring settings in TOML:
```TOML
[plugin.some_plugin_name]
option=value
[component.some_component]
option=value
```
- Add the `txtorcon` optional dependency, used with `tor` extra.
- Add settings for Tor.
- Use Tor endpoint when an `.onion` XMPP domain is used, of when Tor is explicitely
enabled in conf.
- Disable WebRTC, for now it's the safer option as WebRTC can leak IP. It should be
possible to use it over Tor by forcing the use of a .onion relay, but this is to be
explored later (A/V should be to slow to be usable, but that stay relevant for file
transfer or other WebRTC powered things).
- A new `treq` instance using Tor WebAgent is used when Tor is enabled. Properties has
been added to SatXMPPEntity to conveniently get the right instance.
- Another `tor_mode` property has been added with corresponding bridge methods, to see if
Tor is activated and in which mode for a given profile. For now there are only two modes
(disabled, and enabled with WebRTC disabled), but in the future it may be possible to
activate WebRTC over Tor, or to use Tor only to connect to XMPP server, while using
clearnet for other connections.
- Tor dedicated models have been added, and bridge methods have been added to build and
get them.
- Fix tests.
rel 473
This is a huge configuration refactoring.
The configuration system now uses Pydantic settings, with better validation, better
support for environment variables and other goodies.
The main config files now use TOML instead of INI, this allow for better handling of
complex types, notably dict.
Plugins, components and frontends can declare their config settings in a class which will
be imported dynamically.
`maybe_async` can be used to call a function which may be or not `async`. It doesn't use
`defer.ensureDeferred` contrary to `ad_defered`, so the call stack is not lost, and it's
easier to debug when an Exception is raised.
Autocrypt header must be transmitted in both directions to allow opportunistic end-to-end
encryption with this protocol.
Moved email validation regex to `tools/common/regex.py`, as it can be used in other
locations.
rel 456
This is a series of changes necessary to prepare the implementation of remote control
feature:
- XEP-0166: add a `priority` attribute to `ApplicationData`: this is needed when several
applications are working in a same session, to know which one must be handled first.
Will be used to make Remote Control have precedence over Call content.
- XEP-0166: `_call_plugins` is now async and is not used with `DeferredList` anymore: the
benefit to have methods called in parallels is very low, and it cause a lot of trouble
as we can't predict order. Methods are now called sequentially so workflow can be
predicted.
- XEP-0167: fix `senders` XMPP attribute <=> SDP mapping
- XEP-0234: preflight acceptance key is now `pre-accepted` instead of `file-accepted`, so
the same key can be used with other jingle applications.
- XEP-0167, XEP-0343: move some method to XEP-0167
- XEP-0353: use new `priority` feature to call preflight methods of applications according
to it.
- frontend (webrtc): refactor the sources/sink handling with a more flexible mechanism
based on Pydantic models. It is now possible to have has many Data Channel as necessary,
to have them in addition to A/V streams, to specify manually GStreamer sources and
sinks, etc.
- frontend (webrtc): rework of the pipeline to reduce latency.
- frontend: new `portal_desktop` method. Screenshare portal handling has been moved there,
and RemoteDesktop portal has been added.
- frontend (webrtc): fix `extract_ufrag_pwd` method.
rel 436
Implement XEP-0343: Signaling WebRTC Data Channels in Jingle. The current version of the
XEP (0.3.1) has no implementation and contains some flaws. After discussing this on xsf@,
Daniel (from Conversations) mentioned that they had a sprint with Larma (from Dino) to
work on another version and provided me with this link:
https://gist.github.com/iNPUTmice/6c56f3e948cca517c5fb129016d99e74 . I have used it for my
implementation.
This implementation reuses work done on Jingle A/V call (notably XEP-0176 and XEP-0167
plugins), with adaptations. When used, XEP-0234 will not handle the file itself as it
normally does. This is because WebRTC has several implementations (browser for web
interface, GStreamer for others), and file/data must be handled directly by the frontend.
This is particularly important for web frontends, as the file is not sent from the backend
but from the end-user's browser device.
Among the changes, there are:
- XEP-0343 implementation.
- `file_send` bridge method now use serialised dict as output.
- New `BaseTransportHandler.is_usable` method which get content data and returns a boolean
(default to `True`) to tell if this transport can actually be used in this context (when
we are initiator). Used in webRTC case to see if call data are available.
- Support of `application` media type, and everything necessary to handle data channels.
- Better confirmation message, with file name, size and description when available.
- When file is accepted in preflight, it is specified in following `action_new` signal for
actual file transfer. This way, frontend can avoid the display or 2 confirmation
messages.
- XEP-0166: when not specified, default `content` name is now its index number instead of
a UUID. This follows the behaviour of browsers.
- XEP-0353: better handling of events such as call taken by another device.
- various other updates.
rel 441
`init_script_path` option can be used in `[DEFAULTS]` to run a script at the end of
backend initialisation. A new `init_pre_script` method is used to wait for backend to
reach this stage (designed to be used mostly by CLI frontend), then the usual `ready_get`
method is finished once the script is finished.
