During first scss convertion, classes called as mixins were changed to
an @extend instruction which was the best approximation given the fact
there is no equivalent in sass to do that. The problem is that the
instruction is slowing the scss computation a lot and might also break
the style in unexpected ways because of the complex unwanted selectors
the instruction induces.
This commit removes the need of extends. This is done case per case.
Sometimes this involves adding classes in xml, sometimes to change the
style a little, ... The button rendering refactoring which was made
at the start of the LESS to SASS merge was also done in prevision of
this.
After this commit, sass computation is like 5-6 times faster than less
computation while it was like 10 times *slower* before this commit.
Convert content so that the assets compile on app installation. The
style is still broken after this as the variables/mixins/... are not
defined in the right order (as it did not matter in LESS but does in
SCSS).
This commit basically changes:
- Variables: @var_hello -> $var-hello
- Mixins: .mixin_world() {} -> @mixin mixin-world {}
- Classes used as mixin: .my_class() -> @extend .my_class
- Here there were no other solution than to convert the use of
a mixin call by the use of an extend as a first approximation
- LESS functions -> SCSS functions (e.g. fade -> rgba)
- Move first variable definition before the variable is used
- Still need to make sure last variable definition is at the
right place
Note: the scss version of the lib file was not up to date, I manually
updated it to our current version (very small changes). The lib will
probably have to be changed with bootstrap 4 anyway.
Note 2: the file is left included in the common assets but it would
probably make more sense to put it both in backend and frontend instead.
* test_main_flows
The style of a primary/link button is given by the
'btn-primary'/'btn-link' class. The old 'oe_highlight'/'oe_link'
class should still be supported in views but not rendered in the DOM.
Button rendering was most of the time wrong, incomplete or performed
with duplicated code. This led to:
- Both primary and default classes were put on some statusbar buttons
- Icons were not able to be put for statusbar buttons
- Stat buttons and list buttons code were duplicated
- Classes and style were added twice
- ...
We have a need to be able to abort a pending xmlhttprequest. This is
necessary for the longpoll system so it can restart properly.
With this commit, we simply propagate the abort method from the jquery
deferred to the deferred returned by the session rpc method.
Note that this is not documented on purpose. The reason is that we do
not want to commit on maintaining this feature.
It is kind of awkward to have a deferred which can be aborted, but once
it is chained, the abort feature is lost. So, for now, this will stay a
lower level feature used exclusively for our needs.
Here is a scenario that could happen before this commit:
1. web client is started
2. it binds a callback on hashchange
3. it performs some async work (read initial action for user)
4. the user (or more likely, a tour) changes the url
5. the hashchange handler is called, and it starts a new action
6. the async work for the web client is over, it then executes the
default action and/or clicks on the first menu
7. we are now in the default action state, instead of the desired action
state (the one from the new url)
With this fix, we simply do nothing when there was an hashchange before
the initial action was done.
In the calendar view there is some mixing of moment with:
- a local date with given timezone
- a UTC date with added timezone manually applied
The setDate function of calendar model sometimes received either of the
two types, and always transforms them to UTC date with added timezone.
This cause eg. the following issues:
- in the minicalendar, if we click on a day (eg. 12) the day highlighed
will be the one of 12th at midnight with timezone applied. So with a
negative timezone the 11th would be highlighted in the mini calendar
(but the 12th would still be the day shown).
- if we change scale (week/day/month) or open calendar or use next or
previous button: we call `setDate` with a UTC+timezone moment. So each
time this is done, we get an additional timezone
eg. when opening the calendar view on the 12th at 5:00 with GMT-0200:
-> the target date is 12th at 1:00
-> if we set the day mode: target date is 11th at 23:00
-> if we set the week mode: target date is 11th at 21:00
With this change we keep target_date and highlight_date as local date so
setDate always receives a local date and jquery-ui DatePicker gets the
date in the expected format.
opw-1831368
closes#24188
Issue: when displaying the shortcuts with ALT key, when a dialog is open
the shortcuts are displayed in absolute position above the gray overlay
of the dialog, so it is possible to click on the buttons of the
shortcuts.
This fix will remove this possibility.
Reproduce the issue:
- open the manual reconciliation (it is a client action)
- add a reconciliation (using the + sign)
- expand it (using the > sign)
- put the focus on the first field and hit TAB
--> you would expect that the focus moves to the second field
--> what happens is: the focus stays on the first field
Before this fix, when in a client action (or any place that contains
fields) that contains fields, the keyboard navigation did not occur
correctly because the default behavior of the TAB key was prevented.
After this fix, if the navigation_move event was not stopped, we allow
the default behavior of the TAB key.
In the form view, a new problem appeared since we introduced the extra device
size_class: the buttonbox was folded for large screens, even if there is
only two buttons.
The reason is that the form renderer code used a list to map size class
numbers to the number of buttons allowed to be displayed. With the
extra size class, this number was undefined, which caused an issue.
In the form view, a new problem appeared since we introduced the extra device
size_class: the buttonbox was folded for large screens, even if there is
only two buttons.
The reason is that the form renderer code used a list to map size class
numbers to the number of buttons allowed to be displayed. With the
extra size class, this number was undefined, which caused an issue.
Revision on https://github.com/odoo/odoo/commit/32b8cec5361174c92c45a4975d4e02f1fb448aa2
Before this commit, when we switch channels in the Discuss app, the breadcrumb
was visually not updated: it keeps the name of the first opened channel.
With this commit, the breadcrumb is updated on each change of channel, so that
it displays the name of the currently opened channel.
Technical details:
This is due to a change of logic of updating the GUI of the breadcrumb:
before the refactoring above, discuss had to pass the content of the
breadcrumb to the control panel, in order to update the breadcrumb visually.
After the refactoring, the action manager should warn the control panel to
update the breadcrumb on any change of the title. This is already the case
for `ir.actions.act_window`, but not for all types of client actions.
When Odoo display a phone number number, on phone or if VOIP is
installed a invisible `­` character was added from 9.0 up to
saas-11.2 so an applications such as skype would not be concerned with
the number.
But if someone copy-pasted from this to somewhere else in Odoo, we would
get incorrect number with this invisible caracter. So this change make a
special case for the "FieldPhone" so the character is removed from data
saved.
opw-1834858
closes#24223
Before this rev., a crash occured when a default_get returned a
command 0 (CREATE) for a one2many, with a value for a date(time)
field (e.g. [0, false, {some_date_field: '2018-12-08'}]). This
was because the related values (inside the commands) weren't
parsed. This only impacted date and datetime fields, for which the
parsing converts the string into a moment instance.
Fixes#23672
Have the create_date field displays on a one2many list
Try to add a record onto the list
Before this commit, when the virtual record is added onto the list
it crashes because the datetime field is undefined and cannot be parsed
After this commit, It doesn't crash though the value would have to wait the save on the main record
to be updated (because it is the create_date)
OPW 1834250
closes#24097
Here is the issue that initiated this work: the bank reconciliation tour
sometimes failed in community. The reason was that the discuss client
action seemed to be loaded after the tour started, and interfered with
the second step of the tour.
What happened exactly was the following:
- web client is loaded
- initial action is starting (but not complete) (this is discuss)
- tour manager is loaded and starts the reconciliation tour
- the reconciliation tour changes the url to trigger an hash change
- the web client calls loadState on the action manager
- the action manager perform a doAction with the tour url
- the discuss client action is now ready, and replace the reconciliation
action
- the tour is broken.
The underlying cause is really a generic problem: the action manager
does not add the client action starting process in its 'dropprevious'
concurrency primitive. This is probably not really an issue for the
act_window actions, since they have custom loading code that make sure
the loading process is put in the drop previous, but it is definitely a
problem for all client actions.
The solution is quite simple, and should also protect the act window
renderers: put the action controller in the drop previous
In Less, the size_class XL is now defined and can be used in media-queries.
It is also added in JS for the sake of consistency.
This new size is needed for a fix in enterprise.
Whenever a widget is destroyed, it should be as inactive as possible.
Before this commit, it could happen that a widget was destroyed after
the willstart method, but before the start method. In that case, bad
things could happen.
For example, the discuss client action tries to look up some information
into the session, with the getSession method. This causes a crash
whenever the widget is destroyed.
With this commit, we introduce a new 'fail fast' feature to our qunit
test suite: when it is activated, the qunit test suite will immediately
stop after the first failed test.
It is accessible as a flag in the url (failfast), or by clicking the
checkbox in the UI. It is currently not activated by default.
Note that this commit also change the url for the runbot phantomjs test
in order to activate this feature. This allows us to increase the global
timeout for the js test suite without fear!
In Less, the size_class XL is now defined and can be used in media-queries.
It is also added in JS for the sake of consistency.
This new size is needed for a fix in enterprise.
Whenever a widget is destroyed, it should be as inactive as possible.
Before this commit, it could happen that a widget was destroyed after
the willstart method, but before the start method. In that case, bad
things could happen.
For example, the discuss client action tries to look up some information
into the session, with the getSession method. This causes a crash
whenever the widget is destroyed.
With this commit, we introduce a new 'fail fast' feature to our qunit
test suite: when it is activated, the qunit test suite will immediately
stop after the first failed test.
It is accessible as a flag in the url (failfast), or by clicking the
checkbox in the UI. It is currently not activated by default.
Note that this commit also change the url for the runbot phantomjs test
in order to activate this feature. This allows us to increase the global
timeout for the js test suite without fear!
Here is the issue that initiated this work: the bank reconciliation tour
sometimes failed in community. The reason was that the discuss client
action seemed to be loaded after the tour started, and interfered with
the second step of the tour.
What happened exactly was the following:
- web client is loaded
- initial action is starting (but not complete) (this is discuss)
- tour manager is loaded and starts the reconciliation tour
- the reconciliation tour changes the url to trigger an hash change
- the web client calls loadState on the action manager
- the action manager perform a doAction with the tour url
- the discuss client action is now ready, and replace the reconciliation
action
- the tour is broken.
The underlying cause is really a generic problem: the action manager
does not add the client action starting process in its 'dropprevious'
concurrency primitive. This is probably not really an issue for the
act_window actions, since they have custom loading code that make sure
the loading process is put in the drop previous, but it is definitely a
problem for all client actions.
The solution is quite simple, and should also protect the act window
renderers: put the action controller in the drop previous
Whenever a widget is destroyed, it should be as inactive as possible.
Before this commit, it could happen that a widget was destroyed after
the willstart method, but before the start method. In that case, bad
things could happen.
For example, the discuss client action tries to look up some information
into the session, with the getSession method. This causes a crash
whenever the widget is destroyed.
With this commit, we introduce a new 'fail fast' feature to our qunit
test suite: when it is activated, the qunit test suite will immediately
stop after the first failed test.
It is accessible as a flag in the url (failfast), or by clicking the
checkbox in the UI. It is currently not activated by default.
Note that this commit also change the url for the runbot phantomjs test
in order to activate this feature. This allows us to increase the global
timeout for the js test suite without fear!
Here is the issue that initiated this work: the bank reconciliation tour
sometimes failed in community. The reason was that the discuss client
action seemed to be loaded after the tour started, and interfered with
the second step of the tour.
What happened exactly was the following:
- web client is loaded
- initial action is starting (but not complete) (this is discuss)
- tour manager is loaded and starts the reconciliation tour
- the reconciliation tour changes the url to trigger an hash change
- the web client calls loadState on the action manager
- the action manager perform a doAction with the tour url
- the discuss client action is now ready, and replace the reconciliation
action
- the tour is broken.
The underlying cause is really a generic problem: the action manager
does not add the client action starting process in its 'dropprevious'
concurrency primitive. This is probably not really an issue for the
act_window actions, since they have custom loading code that make sure
the loading process is put in the drop previous, but it is definitely a
problem for all client actions.
The solution is quite simple, and should also protect the act window
renderers: put the action controller in the drop previous
Before this fix, when navigating with LEFT and RIGHT keys on a grouped
kanban view, if there were empty columns in the direction of the
navigation, the focus did not change to the next column.
After this fix, the navigation left and right ignore empty columns and
jump to the first non-empty column.
Before this fix, list view and kanban views are scrolling every time
the user uses the up and down keys.
After this fix the list and kanban views will only scroll when the user
arrives at the boder of the window
Consider the following scenario involving a one2many list (non
editable) inside a form view:
- click on 'Add an item', a model opens,
- fill the fields and click on 'Save & Close',
- reopen the freshly created record,
- maybe make some changes (this is optional, but makes sense),
- click on 'Save & New',
- fill the fields and click on 'Save & Close',
-> the second created record isn't added to the list.
The problem comes from the fact that when the dialog is opened for
the second time (in this case by clicking on a record), the 'save'
handler registers an 'UPDATE' command. While this is correct for
the 'Save & New' click (as we update an existing record), this
isn't for the second record which doesn't exist yet, and thus
requires an 'ADD' command.
This rev. ensures that the two cases are handled in the 'save'
handler given to the dialog.
OPW 1829723
In a kanban grouped by many2one with a progressbar, after emptying a column
by moving all records, the column wasn't correctly updated. This was causing,
in the best case, a record duplication or, in the worst case, a traceback.
This issue comes from the fact that, with a progressbar, a read_group is
performed each time a record is moved. When the group (column) becomes empty,
the read_group won't return this group even though the group still appears in
the kankan. When moving back the record in this empty column, there is a
desynchronization between the kanban controller data and the model.
To fix this issue, an option `keepEmptyGroups` is passed when performing the
mentioned read_group ; this mimics the kanban display that keeps empty groups
displayed and there is no desync anymore.
Fixes#23982
In list view, the carret was not correctly positioned in empty cells (i.e.
handle widget in readonly).
This rev. ensures that the position rules are only applied if the cell is not
empty.
Fixes#23984
- Use 'baseline' to ensure that the priority icon will align to the
parent's baseline
- Align the favorite icon in dashboards
- Align the statusbar's caret in form views
Note: this should normally not be the case by design as for most
values there is enough space for the numbers not to wrap. While
waiting for a better implementation, the white-space:nowrap CSS
rule will enforce the behavior.
Thank you to @Jerther for the original commit.
Closes https://github.com/odoo/odoo/pull/24100
Backported fixes:
1. [IMP] web: GraphView: remove hack to render in DOM
- ref. 166319fa4a
- fully backported.
This commit backports the fix that removes the hack for rendering graphs
(delay in a setTimeout), so that the graph is rendered when it is in the DOM.
2. [IMP] web: restore scroll position
- ref. 4e204ed223
- partially backported: only keeps `on_attach_callback`
and `on_detach_callback` in the abstract_controller
and in the abstract_renderer.
This commit backports the use of the hooks methods `on_attach_callback` and
`on_detach_callback` in abstract controller and renderer, so that the
controller of any view warns the renderer when it has been attached in
(or detach from) the DOM.
Also, this commit adds `on_attach_callback` and `on_detach_callback` for the
dashboard app, so that the graph view is correctly rendered.
Fixes#24092
qsm-odoo