*google_drive
Go to a user form view and click on "Change Password" in the action
menus.
=> traceback `Cannot read property 'constructor' of undefined`
This error is raised because `active_domain: undefined` is in the context
and py_eval can't handle an undefined value.
closesodoo/odoo#48915
X-original-commit: 69694c627be42e3d068e48be14026493628a3a8d
Signed-off-by: Aaron Bohy (aab) <aab@odoo.com>
Signed-off-by: lul-odoo <LucasLefevre@users.noreply.github.com>
Co-authored-by: aab-odoo <aab@odoo.com>
Add a new group allowing administrators to remove the ability for
users to bulk-export data from the database. It's pretty minor as
technically the user can still access the underlying object through
the basic methods but it's still a bit of a roadblock.
Users can export by default.
Task 2170900
closesodoo/odoo#45400
Signed-off-by: Xavier Morel (xmo) <xmo@odoo.com>
After this commit, in the graph view, when the user clicks on parts
of the bar and pie charts, it opens the list view of the concerned
records (like what was already done in the pivot view).
This is the default behavior, but it can be deactivated by setting
the attribute 'disable_linking' to true in the view.
Task 2198461
closesodoo/odoo#47796
Signed-off-by: Aaron Bohy (aab) <aab@odoo.com>
Co-authored-by: Aaron Bohy <aab@odoo.com>
Before this commit, when the duration of an event is "24H * n". The event
is set as allDay. This is strange to have a event that start a 8H and end
at 8H of next days and the event is in the allDay slot.
After this commit, we don't change the allDay flag anymore.
Steps to reproduce:
* Open calendar in week view
* Create a event from 8 -> 16
* Then resize the end of this event from 16 -> 8 of the next day (BUG)
Task ID: 2183249
Before this commit, it was possible to create/move an event in the "all
day" slot of FullCalendar even when the mapping field wasn't set.
With this commit, when the "all day" mapping is not present, the "all
day" slot of FullCalendar becomes hidden.
Ref:
https://fullcalendar.io/docs/allDaySlot
Task ID: 2183249
Due to previous commit, in month view, the flag marking events as "all
day" is no more present and so some logic is not applied anymore. This
results into the wrong end-date.
With this commit we avoid to change the end date of an event in
FullCalendar.
Note that it changes the behavior on drag-and-dropped event in month
view:
* Before both the last and this commit, an event if 1H will become an
"all day" event when it's dropped into another cell.
* After them, the time will be the same as the original event and only
the date will change.
Also, one test was already added in commit odoo/odoo@7af13089df
to check that, in month view, when we move a non-"all day" event to
another day, it can't change into a all day event.
This commit adds a second test to cover the case when the "all day"
mapping is present.
Task ID: 2183249
Before this commit, events were sorted by ID in month view.
Event A occurring after event B would be placed on top of B instead
of below.
Steps to reproduce:
* Go in the calendar app
* Insert a event at 12:00
* Insert a event at 8:00
* Insert a event at 14:00
* Go in view month the events are not sorted (BUG)
In month view, events are forced to be "allday", so the sequence used
can't be based on its time but only on its ID.
To fix this, this commit doesn't force the "allday" flag anymore.
Task ID: 2183249
In `calendar_tests.js`, the tests about creating an event with timezone in
week mode only triggers 7 asserts but they were design to trigger 8 of
them.
After analysis, it looks like in the commit odoo/odoo@369c20a99b
the FullCalendar's UI was redesigned and some CSS rules might have add
some margin preventing a MouseEvent from being triggered correctly,
resulting in misadapted tests.
This commit restores the original behavior by adding some pixels to
these offsets, allowing the events to be correctly dispatched and
adapting again the tests.
Task ID: 2092550
This commit updates the FullCalendar library from v3 to v4.
The purpose of this update for us is to add a proper support of the
drag-and-drop of events on mobile.
But also to fix scrolling issues present mainly on smaller screens. In
those configurations multi/inner-scrolls were present resulting into
either the inability to properly scroll the desired content or even
blocking alltogether the scroll.
As a nice-to-have feature, the version 4 removes its dependency on the
jQuery library, allowing a faster loading of a bigger set of events.
Last but not least, this version also reduces its dependency on the
Moment.js library (but still keeps a binding to it for compatibility
purpose).
This commit focus only on the update of the library itself and the
adaptation of the related code to keep it working. Further
enhancements/fixes will be done later on.
To do so it modifies the following points:
* Update the library in itself (js, css...).
* Convert all FullCalendar's method calls to the new syntax (e.g.
`$calendar.fullCalendar('destroy')` becomes `calendar.destroy()`).
* Add a utility method to convert v4 events into the old v3 event
structure to ease the transition from one version to the other.
* Wrap all event's custom fields into a new event's
`extendedProps` object (as its the new way of storing them instead of
directly as a prop. of the event).
* Adapt `slotLabelFormat` and `weekNumberCalculation`.
* Use the view's new names:
** `agendaDay` -> `timeGridDay`
** `agendaWeek` -> `timeGridWeek`
** `month` -> `dayGridMonth`
* Use the option's new names:
** `viewRender` -> `datesRender`
** `eventMouseover` -> `eventMouseEnter`
** `eventMouseout` -> `eventMouseLeave`
** `isRTL` -> `dir`
** `selectHelper` -> `selectMirror`
** `weekNumberTitle` -> `weekLabel`
Ref:
https://fullcalendar.io/docs/upgrading-from-v3
Task ID: 2092550
This commit does 4 things:
- adds a file for deprecated legacy fields.
- moves legacy FieldBoolean into this file.
- converts FieldBoolean to owl component.
- adds a new xml file for component fields' templates.
task id: 2193996
* account
This commit adds a new field registry for component fields.
This field registry is used to only register the new field components
and the other one should register the legacy fields.
Those two registries will be used together while all legacy fields are
not converted to owl component.
As the legacy registry now throws an error when adding a component
field the existing component fields (account.ShowGroupedList and
account.ShowResequenceRenderer) are moved into the new registry.
This commit also refactors some field managers (form renderer, list
editable renderer, calendar popover) to use the new registry and thus
support the component fields.
task id: 2193996
... when toggling an optional field.
This bug wasn't really important until now, but with the new
selection mechanism, it could lead to weird situations (e.g. the
selection box says that all records are selected, but the thead
checkbox isn't checked).
Part of task 2185145
We don't change the multi edit mechanism for now: it still only
applies to the selected records, limited to the current page.
However, we add the information in the confirm dialog to make it
clear for the user.
Part of task 2185145
The export feature already exported the whole domain when the
thead selector checkbox was checked (unlike all other actions
which were executed on the current page only).
Following the new selection feature, this commit uniformizes the
behavior of the export with other actions: if the page is selected,
only the records on that page are exported. To export all records
matching the current domain, the user has to click on the 'Select
all' link in the control panel.
Part of task 2185145
This commit allows the user to apply actions to all records
matching the current domain in list views. It also ensures that
the user is aware of the set of records actually selected.
The number of selected records is now displayed in the control
panel, and the user can, with an additional click, select all
records matching the current domain instead of the ones of the
current page.
Task 2185145
added test case for text paste in autocomplete in the searchbar using mouse,
it should populate the autocomplete dropdown.
task-2204174
closesodoo/odoo#48251
X-original-commit: 010961ee3eac55a1e885ce918b597943078324f6
Signed-off-by: Aaron Bohy (aab) <aab@odoo.com>
Before this commit, coming back on a view when in 'debug=assets' would erase the
previously loaded favorites with an 'undefined' value. This was due to an incorrect
cache invalidation mechanism in the DataManager's load_views' method.
Now the views and filters caches are properly loaded if any of them has been invalidated.
closesodoo/odoo#48085
Signed-off-by: Aaron Bohy (aab) <aab@odoo.com>
before this commit, email widget was not trimming value and due to which
space aroung email value remain as it is added by user.
after this commit, email widget will trim the value and remove leading/trailing
space.
task-2197432
closesodoo/odoo#47470
Signed-off-by: Aaron Bohy (aab) <aab@odoo.com>
before this commit:
when manually adding field in export wizard, string has extra space in
begining.
after this commit:
manually added field in export field list doesn't contain any extra
spaces.
task-2065433
closesodoo/odoo#36759
Signed-off-by: Xavier Morel (xmo) <xmo@odoo.com>
Purpose Of the task is, we need a visual cue on those fields to let
the user know that their value is company-specific.
So, in this commit On company depedent field we added a building
icon with title 'Values set here are company-specific' next to the
label in the form view.
closes odoo/odoo#47178
Taskid: 2001748
Closes: #47178
Signed-off-by: Yannick Tivisse (yti) <yti@odoo.com>
Let's assume a form view with an x2many list, containing itself
an x2many field (no widget, so simply displaying the number of
records). When a record of the list is clicked, a form view
opens and displays the same sub x2many field, with widget
many2many_tags. Both list and form views of the x2many aren't
inline (thus have to be fetched on demand).
Before this commit, it crashed when the user clicked on a row
of the x2many list.
The bug has been introduced by a9cebd5a, which wrongly removed the
code handling this specific situation. At least, now we have a
test attesting that this code isn't useless.
Note that the diff isn't as scary as it looks like (we just added
an if/else, and thus indented the former code inside the else
clause).
closesodoo/odoo#48116
Signed-off-by: Géry Debongnie (ged) <ged@openerp.com>
Before this commit, if an action context would contain a key/value pair
like
search_default_foo : 0
the filter foo would be activated, while it should be not considered
as default.
The present commit fixes that situation.
closesodoo/odoo#47958
Signed-off-by: Aaron Bohy (aab) <aab@odoo.com>
We have an editable list with 4 fields:
- fieldA (say boolean)
- fieldB with readonly modifier, readonly iff fieldA is true
- fieldC with readonly modifier, readonly iff fieldA is false
- fieldD
Add a new row to the list: fieldA being false by default, fieldC
is readonly whereas fieldB isn't.
Change the value of fieldA: fieldB becomes readonly, and fieldC is
now editable.
From that point, before this commit, clicking on fieldC activated
fieldD (instead of fieldC).
The bug occurred since 3fd7b20, which added a delay between
the moment fields are re-rendered (and inserted at the end of
allFieldWidgets) and the moment there previous instances are
destroyed (and the new ones take their place in the list). So when
several fields were re-rendered at the same time, the order in
allFieldWidgets could not match the actual column order anymore,
and cause issues like the one described above.
Part of task 2210323
closesodoo/odoo#48039
X-original-commit: 89b9717c08e148ad94db17d8746faa2312b6a07a
Signed-off-by: Michaël Mattiello <mcm-odoo@users.noreply.github.com>
Signed-off-by: Aaron Bohy (aab) <aab@odoo.com>
When passing a default field chain to ModelFieldSelector, the page data is
not set correctly for the relational field. To understand the issue :
- login to master enterprise all db branch in runbot (debug mode recommended)
- create a mass mailing, add subject, select 'Contact' in `Recipients` field
- click on domain selector and select 'Company' from ModelFieldSelector popover
- click somewhere else to close the popover
- again open domain selector and observe the popover
Current behavior : fields of `Contact` are visible in the page even thogh
`Company` is selected in the ModelFieldSelector
Expected behavior : fields of `Company` should be visible in the dropdown page
This commit fixes the issue by correctly pushing the page data
in ModelFieldSelector.
task - 2058702
closesodoo/odoo#39548
Related: odoo/enterprise#5370
Signed-off-by: Damien Bouvy (dbo) <dbo@odoo.com>
When using a mock environment, some services are started and need to be loaded
for some tests to work as intended. Before this commit, we waited for the result
of the `start` method of each service to be resolved. This was flawed because these
methods are all synchronous, so the awaiting period was actually a single microtask
tick.
To clarify this and still make sure that the asynchronous services have enough time
to load properly, the `Promise.all` has been replaced by an Owl next tick method
(with animation frame request).
Consistently waiting for services to be fully ready does not make sense anyway as
we may want to mock a delay during the starting mechanism.
closesodoo/odoo#47865
Related: odoo/enterprise#9328
Signed-off-by: Aaron Bohy (aab) <aab@odoo.com>
Before this commit, the 'mounted' hook of the control panel in the
report client action was never called, because of a missing call
to '_super'. As a consequence, the 'Print' button wasn't displayed.
closesodoo/odoo#47940
Signed-off-by: Lucas Perais (lpe) <lpe@odoo.com>
Before this commit, the context sent by an action performed by the action menu
could hold an 'undefined' select_all key, which would cause the context evaluation
to crash.
Now, the select_all key is defaulted to 'false' and does not crash in actions that
do not provide it.
closesodoo/odoo#47745
Signed-off-by: Aaron Bohy (aab) <aab@odoo.com>
In JavaScript, the following computation introduces numerical errors:
```
Math.pow(10, -4) = 0.00009999999999999999
Math.pow(10, -5) = 0.000009999999999999999
```
In the context of `round_decimals`, such errors propagate to
`round_precision`, ultimately leading to inconsistencies between Python
and JS (see #47300 for detailed use case).
Instead of computing `Math.pow(10, decimals)`, we parse the scientific
notation to prevent such inaccuracies.
We also introduce tests in line with the Python tests.
opw-2205092
closesodoo/odoo#47694
X-original-commit: cb471897bfa08d1fee6a9f49e90d9c120e1733fc
Signed-off-by: Nicolas Martinelli (nim) <nim@odoo.com>
Use of the new control panel helpers to increase consistency and change the assertions
according to the new DOM/behaviour (e.g. components removed instead of turning invisible).
Part of task 2196029
Co-authored-by: Mathieu Duckerts-Antoine <dam@odoo.com>
Use of the new helpers in control panel (and subcomponents) tests and addition
of new tests.
Part of task 2196029
Co-authored-by: Mathieu Duckerts-Antoine <dam@odoo.com>
1. Test utils helpers adapted to properly instantiate the new Owl control panel
2. New file 'test_utils_control_panel' to create a control panel and use helpers
3. New helper: createComponent (creates a generic component, see jsdoc)
4. Use of the new helpers in the current tests (more in sibling commits)
5. Helper 'addMockEnvironment' is now asynchronous
6. Owl config mode set to 'dev' in test suite (props validation)
Part of task 2196029
Co-authored-by: Mathieu Duckerts-Antoine <dam@odoo.com>
Now that mobile is not supported anymore in community, the mobile layout of the
control panel has been entirely moved to enterprise. This caused some conflicts
with the community mobile test suite (which does not make sense anyway). The failing
tests and related style sheets have been moved as well.
Part of task 2196029
Co-authored-by: Mathieu Duckerts-Antoine <dam@odoo.com>
General cleaning of some helpers and generic files. The idea is that they are
less likely to be entirely re-written in the near future so this is a way to
keep the code clean, readable and compliant to the current standards. Changes include:
- Use of async/await when possible
- Use of const/let instead of var
- Removal of underscore.js
- Use of native JS instead of jQuery when possible
The logic of the affected code remains the same.
This reverts commit d6b854306c.
The reverted commit introduced issues in patching class methods.
Introducing getters in the patch function is not actually compatible
with the original method patching.
Test for patch method is also introduced to guard patch from future
invalid modification that happened.
A more robust way of patching classes is now being developed.
closesodoo/odoo#47557
Signed-off-by: Aaron Bohy (aab) <aab@odoo.com>
The FieldMany2One allows to quick create record (name_create).
However, sometimes, required fields on the co-model make the
name_create crash. In this case, the FieldMany2one opens a form
view in a dialog to let the user specify those fields, and create
a new record with the 'create' method. This was working fine.
The FieldMany2ManyTags embeds a FieldMany2One to let the user
search for and select or (quick) create tags. In this case, failing
quick creation didn't fallback on the dialog form view, but
displayed a server error instead. This commit fixes this issue.
Bug introduced by 50bf8309f8
Task 2192754
closesodoo/odoo#47183
X-original-commit: 63ab065f3298c13bb19ae1245d7fd2652581b254
Signed-off-by: Aaron Bohy (aab) <aab@odoo.com>
This commit allows to pass a bin_size context to False.
Before this commit it wasn't possible because it was
always overriden.
OPW-2208232
closesodoo/odoo#47491
X-original-commit: 17d53cea1043054dfa9c39b16b430539c4082a9d
Signed-off-by: Nicolas Martinelli (nim) <nim@odoo.com>
Commit ca4eaf4 introduced the data-mobile attribute, used
to specify options to pass when executing an action, only in
mobile (like using a fullscreen modal).
However, we forgot to add the 'isMobile' check, so the option
was passed in desktop as well.
This commit fixes the issue and adds some tests.
Task ID: 2155601
closesodoo/odoo#47333
X-original-commit: 6cb68a53106fa7326a7f818461418437d5a0df77
Signed-off-by: res-odoo <res-odoo@users.noreply.github.com>
Let's assume a form view with a boolean field and a one2many field,
displayed for instance with widget many2many_tags. There is an
onchange on the boolean field, which populates the one2many, e.g.
[[5], [0, 0, {display_name: 'name', other_field: 'value'}]]
Before this commit, only the display_name was sent to the server on
save, because 'other_field' is unknown by the webclient.
Similar situations occurred with one2manys displayed as lists, and
onchanges returning values for fields that aren't in the list.
These situations probably didn't exist in Odoo yet, but a pending
task on website event adds one.
This commit ensures that all values returned by the onchange for
the x2many subrecords are sent back to the server when the user
saves.
closesodoo/odoo#46807
Signed-off-by: Géry Debongnie (ged) <ged@openerp.com>
This commit fixes two issues that occurred in the following
situation: in a form view, have a one2many field F1 displayed as a
list, that contains another one2many field F2, displayed with
widget many2many_tags (in the list), and with a list in the form
(in the dialog). In the dialog, the list of F2 displays a
field, say 'name', which has an onchange that sets 'display_name'.
Moreover, there is an onchange in the (main) form view, that
populates F1 and F2, for instance, it returns something like: {
F1: [[5], [0, 0, {
F2: [[5], [0, 0, {
display_name: 'xxx',
name: 'xxx',
}]]
}]]
}
1) If the user opens a related record in F1, and adds (in the
dialog) a related record in F2, then validates the dialog, the
newly created record in F2 isn't correctly displayed in the
many2many_tags (the display_name is `false`), i.e. the onchange
hasn't been applied.
2) If field 'name' in F2 in the list is required, the new record
can't be saved, even if the user enters a correct value.
In both cases, the reason is that we use the incorrect viewType in
datapoints: as we don't specify that we specifically want 'list',
the 'default' one is taken, and we thus use the fieldsInfo of the
many2many_tags, which only knows field 'display_name'. As a
consequence, onchanges aren't correctly applied (issue 1), and
fields aren't correctly reset (issue 2).
Issue reported on task 2189529
When x2many fields are displayed with widgets like many2many_tags,
there is no limit on the number of related records to fetch and
display. In this case, the 'limit' attribute on the datapoints is
undefined. This could lead to weird issues when the limit is used
in computation, e.g.
var index = list.offset + list.limit; // = NaN
There are several occurrences of the above examples in the code,
which can be observed in specific scenarios, e.g. the one encoded
in the test. In this scenario, when the bug occurs, new records
added to editable list views are inserted on top (even if the list
is editable="bottom"), but the edited row is the last one.
Issue reported on task 2189529
Let's assume the following scenario in a form view:
- have a one2many field, say fieldA, displayed as a list,
containing another one2many, fieldB, (no widget, thus
displaying 'n record(s)')
- the one2many list is not editable, so there is a sub form view,
displaying fieldB as a list, and in this list some random field,
say fieldC, is displayed
- have a random field on the main form view, with an onchange to
populate the one2many
- set that random field s.t. the onchange returns something like
fieldA: [[5], [0, 0, {fieldB: [[0, 0, {fieldC: 'value'}]]}]]
- there is now a record in fieldA's list, displaying '1 record'
as value for fieldB
- click on that record to open it in a form view (dialog)
- in the dialog, in fieldB's list, we expect to have a single row
displaying 'value' as value for fieldC
Before this rev., it crashed when opening the record in the dialog,
whether the form view was inline or not.
The crash occured because fieldB was already in the list view, so
datapoints already existed for it (a list datapoint, and record
datapoints for records in the relation, in our example one record
datapoint). However, those datapoints didn't have the fieldsInfo of
the form view. When opening the form view, we added the fieldsInfo
of the form to the datapoint of the record we opened, but we didn't
recursively apply the fieldsInfo to its children. As a consequence,
when rendering the list view for fieldB in the dialog, we haven't
the information about fieldC, and it crashed.
Note that if fieldB wasn't present in fieldA's list view, it worked
fine because datapoints didn't exist before we opened the record in
the dialog.
Task 2120235
How to reproduce
================
You must have
- a form view with a one2many list
- in this one2many list add a column of a reference field
How to reproduce
1. open the record associated to the reference field
2. edit this record in the form modal view
3. click on save
-> Traceback
Bug
===
In `basic_model` the attribute `localData` become inconsistent
because of the method `_fetchReferenceData`.
We want to store the ID of the parent and not the all dict.
closesodoo/odoo#47171
X-original-commit: af411b866aa2052c01168342f8f05ae3dabad83e
Signed-off-by: Aaron Bohy (aab) <aab@odoo.com>
This commit trims leading and trailing spaces in search terms of
many2one.
task-2187414
closesodoo/odoo#45282
Signed-off-by: Aaron Bohy (aab) <aab@odoo.com>
