Sink handling of JS logging, exceptions and websocket timeouts so
calls other than _wait_code_ok handle them somewhat properly: the
issue fixed by odoo/odoo#41231 passed because it occurred during
module loading, which happens during initial page loading (browser_js
> navigate_to > _websocket_wait_event), which ignored logs (and
exceptions though here it's a console.error log), and as a result
reported no failure (and would simply miss that specific test as well
as every test following it).
Also since ChromeBrowser treats console.error as an exception,
important messages should be logged atomically. Merge two consecutive
console.error into a single one at the loading of modules so we don't
just get an exception "error while loading foo.bar" without any of the
useful details.
That ChromeBrowser treats console.error as exception is also why the
new method gets a flag (to suppress this behaviour): in the case of
two console.error, upon encountering the first it's treated as an
error so we try to take a screenshot, which goes through the messages
in order to get the screenshot response, which encounters the second
console.error, which gets treated as an exception, which hides the
first error.
Instead, screenshotting (and more generally _websocket_wait_id) should
treat console.error as a regular logging call, probably.
Also run JS tests in debug=assets for easier debugging (ha!) and
improve formatting of exception object when receiving an exception:
* if we can get a description on an `exception` remote object just
print that, it's formatted to show the exception type, message &
traceback
* otherwise format the garbage that is an "ExceptionDetails" object
Followup of a425695e
The terms were back in 12.0
Courtesy of Juan José Scarafía
closesodoo/odoo#41624
X-original-commit: 85d0c7001a997748d7691205bbb8d066597591a5
Signed-off-by: Martin Trigaux (mat) <mat@odoo.com>
Before this commit, when navigating from test environment (e.g.
page reload), it crashed with following error:
`TypeError: Cannot convert undefined or null to object`
This error comes from mocked bus services in tests: even when they
have been destroyed, they handle the window event 'unload'. They
no longer have a parented parent, so `this.call()` returns
`undefined`, hence the crash.
This commit prevents mocked bus services to listen on window 'unload'
event. Tests must always simulate this window event by explicitly
calling the handler.
closesodoo/odoo#41378
X-original-commit: fd08d04f0a24fea52e7b668b90aec48b33d349a3
Signed-off-by: Alexandre Kühn (aku) <aku@odoo.com>
The im_support module was broken in 12.0 after bus refactorings between
11.0 and 12.0.
closesodoo/odoo#38838
X-original-commit: db37e0cedcd0da46bd57686a076a8165e60d6c3b
Signed-off-by: Olivier Dony (odo) <odo@openerp.com>
Co-authored-by: @alexkuhn
Slovenian language, as many others languages, is not present in the
beta/master projects in Transifex.
For some reason, Transiflex removed all current translations, this was
already fixed in 12, but as there are not automatic forward-port for
translations, this is a manual forward-port.
opw-2060055
closesodoo/odoo#36374
Signed-off-by: Nicolas Martinelli (nim) <nim@odoo.com>
Before this commit, there could be an infinite loop when changing
the chat window state of a document chat window.
This is caused when mutiple tabs overwrite the chat window state
with a different value everytime, resulting to an infinite loop.
Accross multiple tabs, `setItem` and `onStorage` are not synchronous.
That's why tabs should communicate through local storage events
instead of actual value in the local storage.
To illustrate the issue, suppose there are 2 tabs (T1 and T2) with
one document chat windows (C) and the following user interactions:
- T1 and T2 both have C open
- on T1, user folds C
- on T2, user folds C
Now let's add an example of local storage (LS) logics that could
apply from above example:
- T1 and T2 both have C open
- on T1, user folds C
[1] T1 writes 'C folded' in LS
- on T2, user folds C
[2] T2 writes 'C folded' in LS
[3] T2 detects 'C folded' in LS (from [1])
=> T2 writes 'C folded' in LS
[4] T1 detects 'C folded' in LS (from [2])
=> T1 writes 'C folded' in LS
[5] T1 detects 'C folded' in LS (from [3])
=> T1 writes 'C folded' in LS
[6] T2 detects 'C folded' in LS (from [4])
=> T2 writes 'C folded' in LS
etc...
This scenario could have been easily prevented by not writing on
local storage when the chat window state has not changed. However,
this wouldn't fix this scenario that also introduces an infinite
loop:
- T1 and T2 both have C open
- on T1, user folds C
[1] T1 writes 'C folded' in LS
- on T2, user folds C
[2] T2 writes 'C folded' in LS
- on T2, user unfolds C
[3] T2 writes 'C unfolded' in LS
[4] T2 detects 'C folded' in LS (from [1])
=> T2 writes 'C folded' in LS
[5] T1 detects 'C unfolded' in LS (from [3])
=> T1 writes 'C unfolded' in LS
[6] T2 detects 'C unfolded' in LS (from [5])
=> T2 writes 'C unfolded' in LS
[7] T1 detects 'C folded' in LS (from [4])
=> T1 writes 'C folded' in LS
[8] T2 detects 'C folded' in LS (from [7])
=> T2 writes 'C folded' in LS
[9] T1 detects 'C unfolded' in LS (from [6])
=> T1 writes 'C folded' in LS
etc...
This commit fixes the infinite loop issue:
- by turning local storage read during cross-tab communication into
local storage event read instead. This should prevent race
conditions based on read/write operations in local, since their
order is not guaranteed accros tabs.
- by isolating document chat window state to their own local
storage entry. This should prevent a tab from notifying a chat
window state that it didn't change, which may also result in
an infinite loop.
Some tests were adapted from the changes above. Also, some tests did
pass thanks to local storage being mocked by a ram storage: the
handler `_onStorage` should have been called, but it didn't because
it was registered on the actual local storage instead of the ram
storage. In order to make tests pass again, the following additional
changes were required:
- Cross-tab bus service now exports the unique tab ID with
`getTabId`.
- document thread entries in local storage now track the tab ID, so
that self-tab ignore their own local storage writes.
Task-Id 2034997
Closes#36177
Multi is the default api for methods, it is not necessary to explicitly
decorate methods with it, adds clutter and most people use it because
they see that the rest of the code uses it.
Done with `find . -type f -name '*.py' | xargs sed -i '/@api.multi/d'`
According to MDN, HTMLMediaElement returns
> A Promise which is resolved when playback has been started, or is
> rejected if for any reason playback cannot be started.
>
> Note: Older browsers may not return a value from play().
Since un-resolved un-handled promises re-raise and trigger the crash
manager, this is extremely noticeable if the browser doesn't support
the media type or is not currently allowed to play
medias (e.g. configured to prevent auto-play).
This just avoids the error, but eventually it might be useful to
display a toast noting the user could allow sounds iff the rejection
reason is NotAllowedError. This fix would probably be implementable /
implemented on older revisions (which might have the issue, just not
as prominent as unhandledrejection is not hooked into the crash
manager).
closesodoo/odoo#33260
Signed-off-by: Xavier Morel (xmo) <xmo@odoo.com>
Before this commit, mail status manager was regularly fetching
`im_status` of partners with `this._rpc()`. This used the main
thread worker, so it may be overloaded because of that.
To mitigate this issue, mail status manager now uses a new route
`/longpolling/im_status` to fetch `im_status` of partners using
the longpolling thread worker.
closesodoo/odoo#33238
Signed-off-by: Olivier Dony (odo) <odo@openerp.com>
When making a poll, an event is registered on each channel. Once
a notification is sent on a channel, only the events of the
notified channel are removed. A pointer to the event is kept
in all other channel the user subscribes to until those channel
are notified.
Since a user usually has many mail.channels but only a few active ones,
a new event and a bunch of pointers are created at each poll and
never removed.
This fix simply remove the pointers to the current event at the
end of a poll.
closesodoo/odoo#31215
Signed-off-by: Toufik Benjaa (tbe) <tbe@odoo.com>
`this._id` is overriden by the crosstab_bus, so the event handlers
weren't unbound at destroy.
Manual forwardport of odoo/odoo#31578, as we need this for another
branch, in which it makes tests fail.
closesodoo/odoo#31579
Signed-off-by: VincentSchippefilt <VincentSchippefilt@users.noreply.github.com>
When embedding the livechat on an external website, we used to make JSONP calls.
As the support of JSONP calls has been dropped, we now use the CORS mechanism
instead.
This commit is a refactoring of im_status management.
This commit also add im_status in two places, near the author in a mail thread and on each suggestion when using mentions.
A service to manage im_status will have multiple benefits here:
-centralise information, avoid to call the server multiple time for the same im_status
-update all im_status at once and keep consistency in display.
With this commit, the im_status updates are now done by rpc call.
Updates where previously made with the bus but this has some drawback,
since the bus will only give the information 50 seconds after the beginning of
the request, in the worst case, we van wait 2*50 seconds to get an update.
More than that, the im_status where only updates for pinned dm_chat. Dm chat
are synchronized cross tabs, making the use of the bus possible for this purpose.
Since we will need to display im_status not linked to dm_chat, the list to update
will be different from tab to tab making the use of bus difficult for this purpose.
Technical notes:
-im_search has been moved from bus to mail addons since it concerns mail.channel
-Update of an im_status should be reflected everywhere in the page.
Since im_status is a rendered template used in multiple widget, we should add
the correct logic to all concerened widget. The current solution is simple:
use a jquery selector to find every place where im_status is rendered.
-we add a new im_status: im_partner. This will indicate that that
the partner has no user linked to him, making it possible to avoid to ask
for status updates for this partner.
-The update will only be done when the tab is focused. (and will be done
asap once the tab get focus back)
Xavier Morel