A new `record_fd` argument can now be used to specify a file descriptor. If set, the
incoming stream will be written there.
For now `record_fd` must be specified on `WebRTC` creation. In the future, it should
be possible to add or remove it it during a call.
Update the pipeline workflow to handle `SourcesStdin`. A `decodebin` is used to autodetect
the stream type, then audio and/or video elements are created and plugged accordingly.
When packets are lost, artefacts appear and the remove stream video quickly becomes mush
(can be reproduced by blocking UDP for few seconds). To recover from this situation, the
video pad is monitored for GAP events, and when they occur, a new keyframe is requested 1s
after the last GAP.
- Each video source (webcam, desktop, muted) has now a specific config, to adapt
framerate, bitrate and keyframe max distance.
- Improve logging.
- Tuned pipeline to make it error resilient and improve bitrate.
`app.sendPostedEvents` was not handling all events (notably not the window system ones),
resulting in side effects such as the image freezing during A/V calls when the mouse was
not moving for a few seconds.
Instead, `app.processEvents` is now used, even though the documentation discourages its
use: in this context, it should be fine, and it's a simple solution.
Buffer and wait for remote description to be set before sending ICE candidate. This avoids
the sending of ICE candidates before the negotiation is done.
- Instructions have been checked and updated. OS packages dependencies have been added for
Arch Linux.
- Added missing package.
- Reordered/refreshed instructions.
- Old/not relevant content has been removed.
- Tested on latest docker images of Debian/Arch/Unbutu.
rel 466
- Arguments are now grouped in a new `PageData` NamedTuple, making it simpler to use, and
easier to extend.
- Page index is automatically calculated, and can be checked in callback to know which
page to show.
- Form send by calling entity is automatically parsed, using original form to instantiate
the right field types.
- A new `DataForm.get_field` method use Generic type as second argument to retrieve the
right field when getting a `submit` form.
- Updated all classes registering Ad-Hoc commands to use the new system and simplify the
code.
ORM was used, resulting in deserialisation/reserialisation of value checked for race
condition during update. In addition to the overhead, this could lead to modification even
if the value didn't change (e.g., because of JSON key order changed).
The raw value is not used to check if the row has not change between `select` and
`update`, making the method more robust.
- Ad-Hoc callbacks now receive only a `client` and a `page_data`.
- `page_data` keep all important data useful to process current page in a namedtuple.
- Submitted form is automatically parsed, so callback just has to check the data from there.
- Page index is automatically handled, so the handling by callback is no longed needed.
- Session data is now splitted in 2 dict: private for `AdHocCommandHandler`, and public
for the callback.
- add Ad-Hoc commands to
* delete an email
* move an email to another mailbox
* change flags on an email
- The email Message-ID is used to identify to email to manipulate, it is used as <message>
ID on XMPP side.
- Add a `vebose` attribute on IMAPClient to see data exchanged with server.
- Refactor IDLE workflow, to fix synchronisation issues when command were emitted quickly.
- When in IDLE mode, and a command is emitted, there is now a 5s delay to restart IDLE
mode, allowing for a series of command to be run without going to IDLE mode between each
command.
- Fix pubsub nodes which were returned on each disco#items query.
fix 465
Pydantic models are now used for Ad-Hoc commands callback data. The `AdHocCommand` model
can (de)serialise the callback by using its import string.
This change has several benefits:
- better data validation.
- the serialisation mechanism will make communication with frontends easier.
- open the door for automatic documentation.
rel 465
`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.
This new model is easier to manipulate and serialize, and use the benefits from Pydantic's
data validation mechanisms. It has method to easily convert to/from Wokkel data form or
to/from a domish.Element.
rel 465
For mailing-list messages:
- convert all text-only messages to XHTML for consistency.
- detect and flag "noisy" content, i.e. content which are not useful and annoying in
forum-like view (full-lenght quoted previous messages, mailing-list signature)
rel 464
- Pubsub relationships are used to set node hierarchy (comments nodes are set as children of
corresponding pubsub nodes).
- XEP-0499 Extended Pubsub Discovery is used to return hierarchy of pubsub nodes and
items.
rel 463
- Implement Pubsub Extended Discovery, and Pydantic models to handle the data easily.
- `client.discoHandler` initialisation has been moved in `core.xmpp` at an earlier stage
of the workflow as it has to be initialised so XEP-0499 could remove the default disco
handler to handle disco itself.
- Removed `Disco handled` trigger point as it was not used.
rel 463
