Before this commit, the keep alive timeout could occur for many websockets at
the same time resulting in a burst of transaction (cursor is open when connecting,
disconnecting a websocket).
This issue was even worse because the cursor was opened even when it was
not needed: we don't need to open a cursor if there is no callback
registered for the lifecycle event (OPEN/CLOSE).
This commit fixes those issues by:
- adding a random delay to the keep alive timeout of every websocket.
- not triggering lifecycle events if no callbacks are registered.
closesodoo/odoo#110127
X-original-commit: b4fb1359021b6c882ece3037c82aaaf65cfc2e1e
Signed-off-by: Julien Castiaux <juc@odoo.com>
Before this commit, the bus service would fail to initialize its worker
when loaded from a different origin than the worker script URL.
Moreover, the websocket URL was assumed to be on the same domain that
the bus service.
This PR fixes both issues:
- When the bus service is loaded from a different origin than the worker
script URL, create a data URL and use "importScripts" inside the worker
since it is not restricted to same origin scripts.
- Pass the websocketURL to the worker during its initialization.
opw-3118394
closesodoo/odoo#109701
X-original-commit: c1a480bb9ec133b9f94a5bb505380216a7f9c4ce
Signed-off-by: Alexandre Kühn (aku) <aku@odoo.com>
Previous behavior: when debugging the websockets worker, source code is always minified.
Now, if user is in `debug=assets` mode, the asset won't be minified and it will be easier to debug.
@moduon MT-1900
closesodoo/odoo#109583
X-original-commit: 6360bbdbdcc6f2c150ceaa9f7fb5b0760e0f8940
Signed-off-by: Julien Castiaux <juc@odoo.com>
Co-authored-by: Julien Castiaux <Julien.castiaux@gmail.com>
Before [1], the bus was started lazily: either as a consequence of
the addition of a channel to listen to or by manually calling the
`startPolling` method.
Before this commit, the websocket would have been started as soon as
the bus service starts which degrades performances.
This PR fixes the issue by re-introducing the same mechanism as before
that is by starting the websocket either by calling manually the `start`
method of the bus service or automatically when adding a channel.
[1]: odoo#75510
closesodoo/odoo#108741
X-original-commit: 800e175c5bc5d8ba23265c69c5c2cab6957fc73a
Signed-off-by: Alexandre Kühn (aku) <aku@odoo.com>
This commit fixes two things:
- The owl compatibility layer adds an `on` and `off` method on owl's
EventBus so that it can keep being used as it used to in owl 1. To do
this, it needs to keep track of the the callbacks that are attached to
it and by whom. To do this it keeps a Map where the "owners" are keys.
When calling `off` to remove event listeners, we remove the callbacks
but did not remove the owner from the Map, causing the EventBus to
always hold a strong reference to any object that was given as the owner
to the `on` method, preventing the item from being garbage collected.
Objects that do this are typically components, and components hold a
reference to their owl application that contains the entire tree of
components. This was particularly problematic in tests where components
are created and destroyed at a rapid pace.
- The mock websocket creates a MessageChannel so that it can mock the
server side of the websocket, but it never closes that MessageChannel's
MessagePorts, which can cause garbage collection to be slow to happen
[1] (in practice it looks like these are basically never garbage
collected unless closed). This commit registers a cleanup to close the
message ports after the execution of the current test.
[1]: https://html.spec.whatwg.org/dev/web-messaging.html#ports-and-garbage-collectionclosesodoo/odoo#108309
X-original-commit: 4facc396a4e8036592a473002253d5a5daf41a19
Signed-off-by: Aaron Bohy (aab) <aab@odoo.com>
Co-authored-by: Aaron Bohy <aab@odoo.com>
*: base_setup, hr_timesheet, mail, partner_autocomplete, web_tour
Start odoo without -d and with a --dbfilter that allows multiple
databases. Via JSON-RPC access the /web/session/authenticate route
providing a non-filtered database and valid credentials. Traceback,
`request.env` is None.
Since httpocalypse the initialization of the ORM (cursor, registry,
environment) is greedy. It means that the connection to the database is
established very early during the request routing or skip altogether in
case no dbname was known at that time. This contrast with prepocalypse
where the various ORM thingies were lazily setup the first time they
were accessed.
This changement has an important implication regarding authentication.
In prepocalypse, thanks to the lazy approache, a cursor/registry/env
would be setup on the database you just login upon using the
`request.env` for the first time. This was very nice in this regard but
had other problems.
Since httpocalypse such operation is no more possible. Devs must
initialize and use their own cursor/registry/env in case they
authenticate on another database than the one `request.cr` is (maybe)
connected to.
The `/web/session/authenticate` controller is an example of such case.
It crates its own cr/registry/environment after authentication. The
problem the controller uses `ir.http.session_info` and that not all
overrides were updated to use `self.env` (=the env created in the web
controller) instead of `request.env` (=the missing env of the request).
closesodoo/odoo#108063
X-original-commit: 7b9bd9d37731fae724dc5d91da656dab70aa9ad4
Related: odoo/enterprise#35012
Signed-off-by: Julien Castiaux <juc@odoo.com>
*: base, account, crm, hr, hr_attendance, test_access_rights
The various public methods of the ORM can be override in other models,
those overrides sometime don't implement the exact same signature as the
original method in the ORM. In this work we sanitize all the overrides
to ensure a better compatibility. The background objective is to make it
possible to call any public method using kwarg: `search(domain=[...])`.
* `search`, the first parameter was renamed from `args` to `domain` in
0e9adf7 but the overrides were not updated.
* `invalidate_models` and `invalidate_recordset`, a new `flush=True`
parameter was introduced in 9c3b9a4 but the overrides were not
updated.
* `update`, there is a clash between the `update` method responsible for
writing on a record and `update` in bus responsible to update the user
presence. The bus method has been renamed so it doesn't clash with the
ORM.
This sanitization comes with a new linter that verifies that all
overrides of BaseModel public methods share a compatible signature. The
linter has been disabled for `create`, `write` and `default_get` as too
many overrides don't respect the signature of BaseModel.
closesodoo/odoo#106999
Related: odoo/enterprise#34991
Signed-off-by: Julien Castiaux <juc@odoo.com>
Odoo sh needs the update channel event to always come first.
This PR ensures it will always be the case.
closesodoo/odoo#107971
X-original-commit: 1f29e7bec160c1fa6930b3ee8b7753150253a7e2
Signed-off-by: Alexandre Kühn (aku) <aku@odoo.com>
Before [1], the bus was started lazily: either as a consequence of
the addition of a channel to listen to or by manually calling the
`startPolling` method.
Before this commit, the websocket would have been started as soon as
the bus service starts which degrades performances.
This PR fixes the issue by re-introducing the same mechanism as before
that is by starting the websocket either by calling manually the `start`
method of the bus service or automatically when adding a channel.
[1]: odoo#75510
closesodoo/odoo#107878
X-original-commit: 5d7deacf54f37f0938b92a3c45c9f1d1325d1a9f
Signed-off-by: Alexandre Kühn (aku) <aku@odoo.com>
Signed-off-by: Stockbauer Matthieu (tsm) <tsm@odoo.com>
Since #106454 the `isOdooFocused` variable is not updated correctly.
This commit fixes this issue.
closesodoo/odoo#106556
X-original-commit: 0e9351e7bfda370a994a4cc38383bf7d5547dac8
Signed-off-by: Alexandre Kühn (aku) <aku@odoo.com>
Signed-off-by: Stockbauer Matthieu (tsm) <tsm@odoo.com>
Before this PR, the presence service would crash on cross-origin
iframes. This is due to the fact that accessing the parent document
of a cross origin iframe is forbidden for security concerns.
This error was introduced when trying to synchronize the website iframe
and the window focus state in #102479.
This PR fixes the issue by catching the error and fallbacking to the old
behavior given than task-3006023 has been fixed by #53143 anyway.
task-3073242
closesodoo/odoo#106499
X-original-commit: 7210c50df89dce68a554c6d685c33ce74e09d30c
Signed-off-by: Stockbauer Matthieu (tsm) <tsm@odoo.com>
Since the iOS 16.1 update, the WKWebView used in the iOS mobile app
crashes when trying to instanciate a `SharedWorker`... even if this
feature was newly added in iOS 16.0.
Safari iOS doesn't seem to be impacted.
This commit works around this issue by disabling the usage of the
`SharedWorker` in the iOS mobile app, using instead the existing
fallback to a regular `Worker`.
Steps to reproduce:
- On a device running iOS 16.1
- Open Odoo in the iOS mobile app
=> Once connected the app freezes with a gray screen
opw-3045539
closesodoo/odoo#104459
X-original-commit: afac382bd711b1a506629742456d91526cc52c53
Signed-off-by: Adrien Dieudonné (adr) <adr@odoo.com>
Signed-off-by: Pierre Paridans (app) <app@odoo.com>
Some of our current bus notification only insert data in the models system. We
can now simplify the way we handle them by using a generic `mail.record/insert`
handler.
closesodoo/odoo#102863
Signed-off-by: Alexandre Kühn (aku) <aku@odoo.com>
Before this PR, the last notification id known by the client
was not reset after restoring the database. This was an issue
since there can be a gap between the last notification and the
one that has been restored with the database. In this scenario,
messages are not received after restoring the database since the
client subscribes to higher notification ids that the ones that are
created.
This PR fixes this issue by defaulting to 0 if the one the client
passed is higher.
closesodoo/odoo#103311
X-original-commit: 5118da53a6e9adb94577a4f16999522a7e1e8c10
Signed-off-by: Sébastien Theys (seb) <seb@odoo.com>
Signed-off-by: Stockbauer Matthieu (tsm) <tsm@odoo.com>
The `ImDispatch` thread uses the `thread.is_alive` method to check
whether or not the thread should be started. The issue is that, this
method will return `False` after the start method has been called, but
before the run method is invoked leading to the RuntimeError: thread
can only be started once.
This commit fixes this issue by suppressing this error in this case.
closesodoo/odoo#103157
X-original-commit: daded0125a4ec2421b9e817aaaa725da2fd96a20
Signed-off-by: Alexandre Kühn (aku) <aku@odoo.com>
Signed-off-by: Stockbauer Matthieu (tsm) <tsm@odoo.com>
Using patcher.start() can easily lead to incorrect cleanup.
-> after a copy paste, patcher is working, but stop is forgotten
-> stop is present, but won't be called if something fails during the
test
This commit add an utility `start(patcher)` to always have the add
cleanup.
Using a standard way to start the patcher with an automated addCleanup
should prevent this kind of mistake. This is why this commit also
replaces all valid patch.start() (followed immediately by a addCleanup)
closesodoo/odoo#102873
X-original-commit: 7d5a193d86316965a0908c65cfacfb607dc3f3ad
Related: odoo/enterprise#32618
Signed-off-by: Christophe Monniez (moc) <moc@odoo.com>
Signed-off-by: Xavier Dollé (xdo) <xdo@odoo.com>
Before this PR, the presence service was relying on window events
blur/focus to determine whether or not odoo was focused. The issue is
that in the website backend, an iframe is used. When this iframe is focused,
the window looses the focus and vice versa.
This commit fixes this issue by using `window.parent.hasFocus` method.
task-3003950
closesodoo/odoo#102518
X-original-commit: ce57d4494f2bc6a1504f8a7bddddca2a3d2b794b
Signed-off-by: Alexandre Kühn (aku) <aku@odoo.com>
Before this commit, the `updateBusPresence` method could have been
called many times in a row, resulting in many `update_presence` messages
being sent through the socket. This commit solves this issue by throttling
this method in order for it to only be called once in `UPDATE_BUS_PRESENCE_DELAY`
seconds.
closesodoo/odoo#102442
X-original-commit: 0b581f5af509040928babf0e892732c0f4c873e5
Signed-off-by: Sébastien Theys (seb) <seb@odoo.com>
*: bus, calendar, hr_holidays, im_livechat, note, rating, snailmail,
website_livechat.
The mail mock server is getting bigger and bigger. As a consequence,
navigating/keeping it up to date is quiet inconvenient.
This commit solves this issue by splitting it by model/controller.
closesodoo/odoo#98374
Related: odoo/enterprise#32314
Signed-off-by: Sébastien Theys (seb) <seb@odoo.com>
Before this PR, a websocket would wait for the server to close
it with a `SessionExpired` close code to be refreshed after the user
logged in/out. Indeed, the server is responsible to check for outdated
sessions on incoming/outgoing messages.
This is problematic: a user logging in with more than one tab opened,
would have to wait to receive its messages. This PR solves this issue
by refreshing the connection immediately in this scenario.
closesodoo/odoo#102420
X-original-commit: 225f80882a130ae63af193adef412e6b993f2c11
Signed-off-by: Sébastien Theys (seb) <seb@odoo.com>
Signed-off-by: Stockbauer Matthieu (tsm) <tsm@odoo.com>
The session used to call this websocket route needs to have been
initiated with a call to /websocket/peek_notifications first.
closesodoo/odoo#102009
X-original-commit: 49aa391b4ef93b2de578c73a1fe852cea32abb43
Signed-off-by: Julien Castiaux <juc@odoo.com>
Before this commit, loosing the websocket connection would
have resulted in a toast being displayed. This toast is seen
as degrading the UX experience and is redundant with the already
existing "Connection Lost" toast. This PR removes it.
closesodoo/odoo#101501
X-original-commit: a26c8aa18ad6c4f52a4526e20d570ca1bc4123f2
Signed-off-by: Sébastien Theys (seb) <seb@odoo.com>
Signed-off-by: Stockbauer Matthieu (tsm) <tsm@odoo.com>
The original goal here is to remove the warning of the google page
speed but it actually is really better to use 'pagehide' as it is fired
more reliably than the 'unload' event and does not prevent Chrome and
Firefox to use the bfcache (browser optimization that enables instant
back and forward navigation).
The persisted property of the pagehide event is checked before
disconnecting from the worker. As for the presence/multi_tab case, it
was chosen to also consider page freezing as it seems more accurate that
what was done before (the main tab could be throttled/frozen which was
an issue).
Some resources to explain why this is better:
- https://web.dev/bfcache/#never-use-the-unload-event
- https://developer.mozilla.org/en-US/docs/Web/API/Window/pagehide_eventclosesodoo/odoo#101309
X-original-commit: e043891d5074ca04ee1bbed421ae5efe5f2f8b88
Signed-off-by: Quentin Smetz (qsm) <qsm@odoo.com>
Co-authored-by: qsm-odoo <qsm@odoo.com>
Debugging websockets is cumbersome, this commit adds debug
messages to the websocket worker in order to ease debugging.
closesodoo/odoo#101172
X-original-commit: 1abb6c45b06d83b602358a58f351e86b27f3f0c7
Signed-off-by: Alexandre Kühn (aku) <aku@odoo.com>
Since [1] when the bus service was adapted to wowl, the assets watchdog
service was imported in the frontend. Because the assets will be the
right ones on the next visited page, the notification it might trigger
is seen as degrading the website UX.
This commit removes the assets watchdog service from the frontend
pages.
Steps to reproduce:
- Visit a public page in a tab - keep it open.
- From a different browser, log in and rebuild assets.
=> Notification appears in visitor's tab.
[1]: https://github.com/odoo/odoo/commit/3ae11c87e8a395219229416c2777bcdcdacd08ac
task-2922959
closesodoo/odoo#101186
X-original-commit: d2efe6220281c00a9a6be154113ff20c0745dde6
Signed-off-by: Romain Derie (rde) <rde@odoo.com>
The `bus_service` uses the `dbuuid` key from session in order to
know when to drop the last notification id from the local storage:
when the db changes.
The issue is that public pages don't fetch the session, this results
in undefined being added into the local storage. The multi tab service
speculates on the fact that the value stored will always be json
parsable.
This commit fixes the issue by taking into account the fact that session
dbuuid can be undefined. Moreover, the multi tab service has been
updated to return the raw value if it is not parsable.
closesodoo/odoo#100897
X-original-commit: ed992ed03456eac6be5fb802578fcc1a0fc7a732
Signed-off-by: Alexandre Kühn (aku) <aku@odoo.com>
Signed-off-by: Stockbauer Matthieu (tsm) <tsm@odoo.com>
Before this PR, the `_mockIrWebsocket__updatePresence` route of the
bus mock server could lead to errors when no im status ids were
provided. This commit fixes this issue by ensuring im status ids
were passed before calling `_mockIrWebsocket__getImStatus` method.
Moreover, old longpolling routes were still present on the bus mock
server. This commit removes these routes.
closesodoo/odoo#100583
Signed-off-by: Alexandre Kühn (aku) <aku@odoo.com>
In order to know whether or not the websocket session is up to date,
the session is saved when opening a websocket connection. Then, for
each incoming/outgoing message, we check if the session still exists
on the file system. When it does not exist anymore, the websocket
connection is refreshed. This ensures the session is always up to date.
Odoo sh proxies the websocket connection and needs a way to tell
when a websocket session is expired. This commit adds the same
check that is done for each incoming websocket message in the
websocket peek route in order to raise a `SessionExpiredException`
when the session is outdated.
This will allow odoo sh to catch this error and to refresh their
websocket connection accordingly.
closesodoo/odoo#100416
Signed-off-by: Julien Castiaux <juc@odoo.com>
Transform the /websocket/update_bus_presence route to type='json'
which makes more sense as a POST than a get, seeing it does things with
a parameter and doesn't return anything.
Also fixes the _update_bus_presence method that leads to access errors
when accessed from HTTP when the user is the public one (only portal/
internal users should update their presences).
Part-of: odoo/odoo#100416
*: bus, hr_holidays.
The `/bus/im_status` route polls the server every minute in order for
the user im_status to be up to date. This commit removes this poll
by sending the im_status on the bus when updating the current user
presence.
Moreover, before [1], the user bus presence was updated on each poll.
When the user didn't poll for 50 seconds, we assumed the user was
disconnected. Since [1], the bus presence is updated each 30 seconds
by the `im_status` service. This is too frequent: there is no need
to update the user presence so often.
In order not to overhelm the server with unnecessary requests, the update
presence interval as well as the delay to be considered disconnected
have been updated: the former from 30 to 60 seconds, the later from
55 to 65 seconds (assuming that a user that has missed an update
presence tick is disconnected).
[1]: odoo/odoo@a5623d2closesodoo/odoo#100249
Signed-off-by: Sébastien Theys (seb) <seb@odoo.com>
Before [1], only string channels were allowed for polling. This
ensured no one could send a server-side channel from the frontend,
this PR restores this behavior.
[1]: odoo/odoo@a5623d2closesodoo/odoo#100309
Signed-off-by: Sébastien Theys (seb) <seb@odoo.com>
*: auth_password_policy, bus, calendar, event, mass_mailing, stock,
survey, web_editor, web_tour, website, website_event, website_forum,
website_mass_mailing, website_sale
This commit is a first step towards a potential deletion of the
assets_common bundle, although that step would need more work, specs and
discussions as some layouts kinda only use the assets_common bundle
(some take the full assets_common but parts of the assets_backend one
for example).
The main goal of this commit is to have the assets_frontend bundle
directly include the "common" files we need. As a first step, this
commit only blindly duplicates them all into assets_frontend (without
removing the potentially useless ones). The goal is to have those
advantages:
- Reaching a frontend page only calls two main JS files (one normal and
one lazy-loaded) instead of 4 (two normals and two lazy-loaded). This
may help reach a better google page speed (which is becoming more and
more strict).
- The frontend CSS is built as one: the common SCSS which was using
bootstrap variables, or even Odoo-based SCSS added by mistake in
common instead of both backend and frontend is now computed with the
right bootstrap customizations. E.g. the tempusdominus datetimepickers
use bootstrap grays... after this PR, they use the right grays as
customized by the user on the website.
It was also chosen to not have a common "sub-asset" which is included in
assets_frontend. Making assets_frontend completely independent makes
sense (as it probably will for other "main" asset bundles): we can focus
on adding the files each layout needs without the need of worrying if it
impacts unrelated layouts. Sub-assets (when not strictly necessary) is
also a source of errors: extending the "main" bundle instead of the
right sub-asset it may use (like it was the case with the sub-assets of
assets_common: _assets_common_scripts and _assets_common_styles as
explained in the previous commit). So this is indeed a small drawback of
not factorizing the code for the inclusion of "common" files in bundles
but it seems more explicit and easier to maintain that way. Note that
adding "common" file is not the most common usecase anyway, apps
generally only need files in backend or frontend.
The __manifest__ declaration will also likely evolve in more and more
uses of wildcards to match entire directories. In the future, adding
"common" web-app files in both backend, frontend and other "main"
bundles could just be about one line duplicated into each bundle.
closesodoo/odoo#100314
Related: odoo/enterprise#31394
Signed-off-by: Romain Derie (rde) <rde@odoo.com>
Before this commit, the `channels subscription after disconnect` bus
test failed randomly. This occured when the websocket reconnection
happened before the first subscription. Indeed, the first subscription
is made only if the resulting channel list differs from the last one
we sent.
This commit fixes the issue by wating for the first subscription
before disconnecting the websocket.
closesodoo/odoo#100391
Signed-off-by: Alexandre Kühn (aku) <aku@odoo.com>
Before [1], the web_editor module was auto installed. After that commit,
the bus module has been added to the web_editor dependencies. Since
the bus module is not auto installed, the web_editor module dependencies
are not met which means the login colors are wrong (blue instead of green).
This commit fixes this error by adding the `auto_install` flag to the
bus module.
[1]: odoo@a5623d24b77f523aeeafbb8c2bfaf27241ef3c8e
closesodoo/odoo#100212
Signed-off-by: Adrien Dieudonné (adr) <adr@odoo.com>
When stopping the server, all cursors are closed. The issue is that
the ImDispatch thread is kept alive until the main thread exits.
This can lead to errors during server stop: the ImDispatch thread
could try to poll an already closed connection thus raising a
`psycopg2.InterfaceError` exception.
This commit solves the issue by hiding the interface error when
it occurs during server stop.
closesodoo/odoo#100002
X-original-commit: 83777078f1be093c218267c8282e8c11f4fb4e8d
Signed-off-by: Julien Castiaux <juc@odoo.com>
Before the websocket were introduced, the last notification id was
persisted in the local storage in order to use it even after a page
reload.
The last notification id is now stored on the `SharedWorker` which
means this information is lost if the last opened tab is reloaded.
This commit fixes this issue by storing this information by the mean of
the localStorage. Since the localStorage is not accessible from the
worker global scope, the bus service will store this information and
relay it to the worker when starting. The worker will now wait to
receive the last notification id before subscribing.
closesodoo/odoo#99621
Related: odoo/enterprise#31124
Signed-off-by: Sébastien Theys (seb) <seb@odoo.com>
The im status service subscribes to the `become_main_tab`,
`no_longer_main_tab` events.
Since the PR improving the multi tab service, the multi tab
itself is not an event target anymore but exposes a bus instance
that should be used to subscribe to those events.
The im status service is still based on the first version (that was
using env.bus) which means the event will never be received. This
commit fixes this issue.
closesodoo/odoo#99635
Signed-off-by: Sébastien Theys (seb) <seb@odoo.com>
The `displays reconnect notification` bus test relies on a `nextTick`
to assert that a notification is present after the websocket
disconnection.
The issue is that we can't be sure that the notification will be
displayed after this tick. Indeed, we need to wait for: the `postMessage`
to reach the bus service, the notification to be added, the notification
to be added in the DOM.
This commit solves this issue, by waiting for the next render after the
`simulateConnectionLost` method to occur.
closesodoo/odoo#99590
Signed-off-by: Sébastien Theys (seb) <seb@odoo.com>
When executing python tests with Odoo opened in the browser, the
websocket connection coming from the browser can lead to issues
with savepoints/rollbacks. Indeed, during test set up, a savepoint
is created. This savepoint is released at the end of each test.
The issue occurs when the websocket connection opens a cursor (and
thus creates a savepoint) after the one created by the test but releases
it after the one created by the test:
- SAVEPOINT TEST
- SAVEPOINT WS
- ROLLBACK TO SAVEPOINT TEST
- SAVEPOINT WS DOES NOT EXIST
In order to solve this issue, let's prevent browsers from opening a
websocket connection during python tests. This does not apply to chrome
headless since remaining threads are awaited before the end of every tour
(which means the websocket connection will be close before releasing the
test cursor).
closesodoo/odoo#99538
Signed-off-by: Julien Castiaux <juc@odoo.com>
Before the websockets were introduced, longpolling coroutines were
sleeping until postgres notify. Each coroutine was then wake up and
notifications were fetched.
Since the websocket introduction, the main loop, responsible for listening
to postgres sends the notifications itself. This means, the postgres loop
is blocked during message fetch/dispatching and notifications are dispatched
in a sequential fashion resulting in a slow message dispatching.
In order to solve this issue, websocket coroutines are now responsible to
fetch/dispatch notifications, letting the main loop free to relay notifications
as they come and allowing notifications to be sent simultaneously.
When instructed to dispatch available notifications, the websocket coroutines
will try to acquire a cursor. Each coroutine will try up to `MAX_TRY_ON_POOL_ERROR`
times, sleeping between each try. If no cursor can be acquired, the connection is
closed with the TRY_LATER` close code.
closesodoo/odoo#98880
Signed-off-by: Antony Lesuisse <al@odoo.com>
* = bus, hr_holidays, test_discuss_full
- use channel member instead of partner for typing
- use channel member instead of partner for all other return values from server
- remove temporary partner hack in livechat and keep public partner
- remove some obsolete convert data
- adapt format methods accordingly
task-2664853
closesodoo/odoo#98923
Related: odoo/enterprise#30760
Signed-off-by: Alexandre Kühn (aku) <aku@odoo.com>
Before this PR, disconnecting the websocket for a keep alive timeout
would skip the terminate method. When the loop would exit, gevent loop
will keep running and use the CPU up to 100%. This commit ensures the
termiante method is called when disconnecting due to a keep alive
timeout and solves this issue.
closesodoo/odoo#99308
Signed-off-by: Xavier Dollé (xdo) <xdo@odoo.com>
The test_user_logout_outgoing_message is undeterministic because
it speculate on the fact that the channel subscription was made
when we try to dispatch notifications. Sometimes it is not and
the websocket to channel map is empty resulting in a StopIteration
exception being raised. Let's wait for the subscribe to occur to
ensure the websocket will be registered before trying to dispatch
the message.
closesodoo/odoo#99204
Signed-off-by: Sébastien Theys (seb) <seb@odoo.com>
The bus test "display reconnect notification" sometimes fails because
the notification is removed before the assertion occurs. Let's block
the start method of the worker to block the reconnection until the
assertion is made.
closesodoo/odoo#99160
Signed-off-by: Alexandre Kühn (aku) <aku@odoo.com>
When a websocket is disconnected, its subscription is removed.
Each channel leads to a set of subscribed websockets. When no
more sockets are listening to a channel, this channel should be
pop from this mapping or the map will keep growing.
closesodoo/odoo#98766
Signed-off-by: Sébastien Theys (seb) <seb@odoo.com>
The websocket client library is used during tests but is not in
requirements.txt. Therefore, imports are usually wrap into a try/except.
An import present in `websocket_rate_limiting` is not wrapped leading to
errors for those who have not the library installed. This PR fixes this
issue by wrapping this import in a try/except block.
closesodoo/odoo#98763
Signed-off-by: Sébastien Theys (seb) <seb@odoo.com>
*: bus, calendar, hr_attendance, iap_mail, im_livechat, project, snailmail,
snailmail_account, survey, web, web_editor, website_livechat.
This commit is part of the websocket integration in Odoo.
This bus service now communicates with a shared worker in order to provide
a single websocket connection for multiple tabs. It is designed to
be used as a websocket except that events are slightly different,
re-connection is handled automatically. If the browser does not support shared
worker (Safari), the service fallback on a simple web worker.
Available events are:
- connect : fired upon a successful connection.
- disconnect : fired upon reception of the websocket close event.
The close code and reason are given to the listeners callback.
- reconnect : fired upon a successful re-connection.
- reconnecting : triggered when the worker starts to try reconnecting.
- notification : fired upon the reception of notifications.
Since multiple tabs are now handled by a worker, the cross_tab bus is no longer
required and has been removed.
Part-of: odoo/odoo#75510
tsm-odoo