Backport of d9fed5381a78c19ce14ddc8b9233f60d53c65453
Before this commit, when opening the calendar view with a specific locale
in the "week" view
the days were translated but the date format was wrong and fell back to english
This was because the translated terms were passed explicitly, but the locale did not
get passed
After this commit, we do what it takes to pass the locale to fullcalendar
and the dates are formatted with the right pattern
Also, there may be a bug in fullcalendar, because just passing the locale in the options
won't work, it should be instanciated first in fullcalendar's "locales cache"
OPW 1922092
OPW 1934127
closesodoo/odoo#30909
Through many design refactorings, the community caret position was
broken while the enterprise was not. This commit removes the breaking
rule to put it in enterprise only.
closesodoo/odoo#30799
Commit 9af941428880af8e1810dfd9feef494d7a745e98
made it so that the value_from function depends on a variable defined after
the function complete has been called, without any guarantee that this function
would be called after. Apparently this is the right OOP thing to do.
opw 1930523
closesodoo/odoo#30712
- Set the OS in a timezone such as the current day is different from the
day in UTC (e.g. America/Nome before 10:00 AM or Australia/Melbourne
after 3:00 PM)
- Open any datepicker
- The 'little triangle' indicating the current day is wrongly set (one
day before or after)
Knowing that Odoo always creates momentjs date and datetime with the
`UTC` flag set to `true`, the `bootstrap-datetimepicker` does something
which seems inconsistent.
First, it retrieves the `viewDate`, and sets it to the beginning of the
month and week in:
https://github.com/odoo/odoo/blob/1c6c504215f3ef09e6336c92c9d350e87599eaa1/addons/web/static/lib/bootstrap-datetimepicker/src/js/bootstrap-datetimepicker.js#L725
In this part, it is important to note that each `startOf` functions
called sets the hours/minutes/seconds to zero. It means that the
reference time is changed.
Then, it iterates on this newly created date, and determines `today` by
comparing it to `getMoment()` in:
https://github.com/odoo/odoo/blob/1c6c504215f3ef09e6336c92c9d350e87599eaa1/addons/web/static/lib/bootstrap-datetimepicker/src/js/bootstrap-datetimepicker.js#L748
However, `getMoment()` returns the current date and time, but with the
`UTC` flag set to `false`.
Therefore, we compare a UTC datetime on which the reference time has
been changed to a non-UTC datetime, which fails to give the appropriate
current day.
There are two approaches to solve this. The first possibility is to
change the way Odoo defines momentjs dates and datetimes, maybe by
removing the `UTC` flag at creation. This sounds like a bad idea, since
other widgets or views (such as the calendar or the pivot view) make use
of them. This is likely to introduce a bunch of new issues with TZ in
these views. The second approach is patching the library to fit our use.
Although we usually don't do such a thing, this allows to specifically
solve this use case, and in particular placing the 'small triangle' at
the appropriate date without impacting any other part of the system or
the library. It can be easily performed by comparing the dates and the
months to make it work.
opw-1915251
closesodoo/odoo#30538
Before this commit, the web client had a naive strategy to handle lost
connections: it tried to poll the server every 2 seconds until a rpc
succeeds.
This works quite well from the perspective of the user, but may be a problem
from the perspective of the server. If a server is down for a longish period,
then each users active tabs will then perform a request every 2 seconds. This
means that the server will be progressively hammered by many requests, which
will clutter the logs, and make it more difficult to gracefully recover.
With this commit, we simply exponentially increase the delay each time, and add
a little jitter to give a better distribution.
Cherry-pick of 4a3f04bcc5closesodoo/odoo#30136closesodoo/odoo#30596
For editing many2one fields (and partially many2many) most widget show
an autocompleting list of targeted records.
They thus have `autocomplete="off"` to prevent browser completion.
But chromium has an history of breaking `autocomplete="off"`, see:
- https://caniuse.com/#search=autocomplete
- https://crbug.com/468153
- https://crbug.com/587466
- https://crbug.com/914451
- https://crbug.com/923895
It seems that since chromium 71, the heuristic to ignore
`autocomplete="off"` has become more aggressive and for example if there
is at least 3 fields like an address in a page, chromium will ignore
`autocomplete="off"` for the fields like an address.
So for example the eidting the many2One field with placeholder "Country"
in a contact page now has a browser autocomplete menu that is:
- hidding the many2one autocomplete
- going to save empty country it appeared visually filled if the
autocomplete result was selected.
With this changeset, the placeholder in the many2one instance is
interspersed with U+FEFF charcters (ZERO WIDTH NO-BREAK SPACE) so the
browser does enable the autocomplete feature by force.
This should thus remove the issue (until it is fixed by chromium) in the
case of field named "Country" or matching other regexes in this file:
https://github.com/chromium/chromium/blob/cdb1b2073f12/components/autofill/core/common/autofill_regex_constants.cc
U+FEFF has been chosen instead of more recommended characters because
other have been shown erroneous for printing in some windows
configuration (see cb2a3afa7).
10.0 version of #30439
opw-1930588
closes#30439closes#30449
Before this commit, the menu to go into the backend from the website was
not possible to open on iOS without opening the right menu first. This
was because position: fixed is not able to counter overflow: hidden in
iOS. This is the fix for stable versions, the proper solution will be
done in master.
Closes https://github.com/odoo/odoo/issues/29390
task-1923151
closesodoo/odoo#30507
Before this commit, in the filter, when using a custom filter with a
date field, the operators "is after" and "is before" were set to execute
"is after or equal" and "if before or equal" respectively. That means it
includes in the results, records with the same date as the filter.
Now, the operators 'is after' and 'is before' won't include the records
with the same date as the filter. And the operators 'is after or equal'
and 'is before or equal' will include the records with the same date as
the filter.
Fixes#29868
OPW-1921756
closesodoo/odoo#30560
Open a pivot view with default filters automatically activating
Disable those filters
Change the groupBys on the rows and cols of the pivot
Save your result as a favorite filter
Before this commit, the filter did not contain the context information for columns groupby
Also, the search view did not exploit the context of the new filter as it should be
After this commit, the flow works as expected
OPW 1913732
closesodoo/odoo#29415
Before this fix, in studio when the user select the field then focus out
then focus in and selected a child field, an exception is triggered because
the last page is visible but the data are removed.
backport of 12.0's 9eef25ae0a
opw-1928453
closes#30277
Before this rev., a crash might occur when the user quickly
switched twice between pages (e.g. go to page 2, then page 3) on a
slow network.
In the given example, when data of page 2 returned, the list was
re-rendered. Unfortunately, the offset of that list' datapoint was
already changed due to the switch to page 3, meaning that the view
tried to render a page that wasn't loaded yet, leading to a crash
if there were modifiers to evaluate, or to empty records being
displayed.
This rev. ensures that the view is rendered with the data of the
page it expects.
closesodoo/odoo#30150
Start on the modal obtained by a "search more". The offset is never reset.
So suppose you are on page 2, looking at record 81-160.
Do a research that gives less than 80 records.
The result of the search is nothing, since is has been done with a 80 offset.
It should be reset to 0 when we do a new search.
opw 1920826
closesodoo/odoo#30109
Most operations on X2Many widget possibly propagates a field context
(name_create, name_search, ...) but the original read or read on an
onchange does not.
With this changeset, the field context is also used in these instances.
The assertions added in the added test failed with:
expected: ["world"], result: [undefined]
expected: ["world","world"], result: [undefined,undefined]
fixes#29203
opw-1914466
closes#29866
Fine-tuning of commit e7fab23420
In a one2many with more than one page with a limit of k records, add a line l.
It increments the list limit to k+1. Remove a line above (of index <= k).
It will try to load a line l' to replace it, that it will find on page 2.
However when deciding if it should read missing fields on that record, it will
scan at most u records, skipping virtual ids.
Here u is the min between the current number of lines and the list limit
without the temporary increase due to the added record.
Since there are at least two pages, u = k.
However since l is in the list at index k, then l' won't be read, and since it
was on page 2 its data hasn't been loaded.
Therefore the js would traceback when trying to evaluate it.
opw 1913361
closesodoo/odoo#29338
Use case: a x2m (ex: attribute values) displayed in a x2m (ex: attributes on
product) using different fields in different views (ex: tags in product form view
and list in product attribute form view). When opening the x2m record (in a
wizard) then going back on the form view, the tags are empty because the
`display_name` value has been lost when fetching x2m for the second time.
In this case, the list `fieldsInfo` needs to be updated with the existing
(default for tags) one so all the fields will be correctly loaded.
Task 1916891
closesodoo/odoo#29598
When filtering the calendar view on search filters, the filter `context`
was not taken into account during the reload.
For example, in the Calendar app, the filter 'My Meetings' has a `context`
{'mymeetings': 1} which was not passed to `search_read`.
Task 1917628
closesodoo/odoo#29531
When activating a filter on a list view,
the environment of the action contained a context with an orderedBy key
which was then used by every controller and model in the various
parallel flows, i.e. changing actions
After this commit, we prevent it by excluding the key when executing an action
OPW 1913190, 1914019
closesodoo/odoo#29186
Recent works in the orm (see [1] for example) changed the behaviour of
the server in the case of onchanges: it tries to optimize Odoo by
sending a minimal diff to the web client in onchanges.
Concretely, this means that the web client receives commands 4 (link to)
instead of command 1 (update + record data). It also sends a minimal
diff in that case. This is a big issue for the web client, because it relied
on the knowledge of the data from the command 1 to be able to evaluate some
modifiers.
The good news is that we can mitigate the issue in one case: we do not
actually need to evaluate the modifiers in the case of a command 4
(comming from the server), because we can assume that the record is
valid. So, we simply bypass the check in that case.
Note that this should fix most of the problem, but a deeper problem
still remains: a crash can still happen when an onchange modifies a
record in a one2many (but in a different page), and we also have a
modifier which needs to be evaluated.
opw: #1904514
[1] https://github.com/odoo/odoo/pull/29086closesodoo/odoo#29172closesodoo/odoo#29086
Have:
- Model A with a x2m to model B
- Object A with multiple objects B in x2m (that will be sorted)
- On the list view of model A, click on a column to sort by a field
- this field shall not exist in B
- open a record A (towards form view)
Before this commit, the sorting of the x2m crashed, because the key orderedBy was propagated
Due to commit e713ed2a69
After this commit, it doesn't crash
OPW 1913133
closesodoo/odoo#29094
Have a form record with a progress bar
Modify the max_value of the progress, as well as the current_value bar through an onchange
Before this commit, the new max_value was not taken into account
After this commit, it is.
closesodoo/odoo#29056
When creating a SO line and clicking on search more for product_id
Before this commit, every user had the button "Create" to create a product.product
After this commit, only user in groups that actually can create a product have the button
OPW 1910371
closesodoo/odoo#28924
Define a x2m tree view with a default_sort including more than 1 field to order the list against
Before this commit, the second field was never used
After this commit, it is used
OPW 1909785
closesodoo/odoo#28840
Put the browser in timezone Sao Paulo (Or like West of UTC)
Trigger the autocomplete (and the subsequent search) on a date field
by writing a date (in the locale format) in the search view
Before this commit, the domain sent to the server contained the date as the day
**before** the one asked for in the input
This is because, the string input is parsed and gives:
input = 12/02/2018
When creating the moment object:
> since there is no explicit time, moment will interpret it as 12am (midnight)
> We force moment to consider the string as being UTC (function: moment.utc())
> the moment contains, as output the time ** 2018-02-12T00:00:00 UTC **
When getting the facet value for making the domain
> we call toDate on the moment object, which, according to the browser is UTC and will
be converted to the locale timezone before formatting
> And the domain will use the date 2018-02-11T22:00:00 Brazil/SaoPaulo
which gives the day before the one we asked for
After this commit, this issue doesn't arise, because we use the string representation of
the moment object in search_inputs (i.e 2018-02-12T00:00:00 UTC), to create a new one
*BUT* we create it with a date format that excludes the time from being interpreted.
Also, the resulting moment object in this case is not flagged as UTC anymore
OPW 1903224
related to: #28446
On a pure list view, click on a column to sort the records by that field.
Save the filter as favorite
Reload, now apply the filter
Before this commit, there were two issues:
1. The order of the list was never propagated to search view, which couldn't save
the filter with its attribute "sort"
2. There was no mechanism to load the attribute sort from a loading Filter
After this commit, those two issues are corrected
Unfortunately, complex search view flows are not testable in v11.0
OPW 1906968
closesodoo/odoo#28766
In the case of a One2many list view in a `<group>` tag, the 'External
Link' button of a Many2one field (which allows opening the form view) is
not visible.
The `flex: 1 0 auto;` should not be applied in the list view.
Corresponding PR in v12: #28741
opw-1903158
closesodoo/odoo#28812
Steps:
- open a modal form view from tree view first line and make a change
- open a modal from the second line and make a change
- open the second line back: the change were lost
In a given number of conditions (a many2one inside the form view, there
is a one2many in the form view not inside the tree view, ....) this
happened because:
- when we save the first line, we get onchange data for the second one
- this onchange data can't be applied before second line is loaded, but
it is saved so it can be applied if second line is ever opened
- when the second line is opened the saved change is applied => OK
- when the second line is opened a second time the saved change is
applied again, possibly overwriting posterious change => BAD
With this changeset, when the record with unapplied onchange is loaded
for the first time, the unapplied onchange is applied only once.
Added test without the change failed with:
- only 1 turtle for second partner (result: 0 not 1)
- second partner turtle is Michelangelo (result: "" not "Michelangelo")
opw-1846820
closes#28644
On a form view (of model A), define a many2one (model B) with a [form|tree]_view_ref context key
In that view, have a x2m field (model C) with a different [form|tree]_view_ref context key
Before this commit, the second load_views (i.e. on the model C) crashed because
the wrong context key was sent to the server, which returned the view to which the _view_ref points
i.e. the view of the model B
After this commit, we clean the context of subviews loading, because those keys are only useful for the python
and one shot. Hence, there is no crash
OPW 1903780
closesodoo/odoo#28260
Lucas Perais (lpe)