When using another decimal separator than `.` (dot)
in the language settings,
it wasn't possible to build an advanced search specifying
the decimals.
e.g. with a language with decimal separator `,` and thousand
separator `.`,
if you want to search invoices with amount total 366,38
The advanced search "Total" "is equal to" obliged you
to enter your number with `.` as separator (366.38),
and then, when entering the search, the `.` was
regarded as the thousand separator, giving as domain
`('amount_total', '=', 36638)`, which is not what you asked.
opw-634201
PS: The `|| '.'` in the xml template are only for
retro-compatibility, so if the server sources are
updated, but the browser cache is not refreshed
(meaning the Javascript code isn't refreshed, and,
therefore, `widget.decimal_point is undefined)
it still works.
In the search input, the result of filtered search with an arrow was
placed under the currently selected arrow.
Hence if we get to another arrow filter before the result is received,
the result is added under the wrong arrow.
This fix adds the result under the right arrow.
closes#6442
opw-632519
When a many2one field of a searchview was selected
by default, through a default_*_id within the context,
the many2one value name wasn't translated.
e.g. with Spanish loaded (and l10n_multilang installed),
translate a project.project name in Spanish.
Then, while being in Spanish, in the project.project kanban,
click on the Tasks link of a project (tareas),
then, notice the value of the project name in the
search bar.
opw-632818
Currently (since commit e6a2d222d6), the domain of a search view field tag
is dropped.
This previous change was done since some of these domains refers to a
field not available in the javascript.
This fix drop the domain only if it can't be evaluated in the javascript
side.
opw-632030
This patch removes the asynchronous code that was
still used to open the autocompletion drawer.
Even with the setTimeout delay forced to 0, the
asynchronous handling meant the final ENTER
keyUp event could be processed before the opening
of the autocompletion drawer related to the last
key pressed. This would trigger the search with
the search string of the previous autocompletion
popup, missing the last few keys.
Also handle ENTER at keyDown, because in some
rare cases a barcode scanner may emit only
the keyDown event for ENTER, without the
corresponding keyUp, causing a merge of two
successive scanned barcodes.
Based on many tests conducted with different
barcode scanners, the only thing that
is guaranteed is the correct order of the keyPress
events for all regular characters, and the
fact that the ENTER keyDown and/or keyUp will
always be sent after them.
All other events can be mixed or simply missing
if you make a rapid succession of scans, especially
if done with a long series of barcodes.
Scanners tested (with 1ms data transmission delay):
- Honeywell Eclipse MS5145
- Dacomex Slim 50mm CCD scanner
This patch hopefully solves an annoying issue with the search view:
when the user types really fast (or uses a barcode scanner for example),
the resulting search string was (sometimes) missing the last few
characters. I'm unclear on the exact reason why it happens. From the
code, I can only guess that it happens because the scanner use a TAB
key instead of ENTER (TAB is handled in keydown event, ENTER in keyup),
or that the key events aren't in a correct order (if the user press a
key A, then press ENTER, then release ENTER, then release A, it results
in an empty key).
Anyway, the way this patch probably solves the issue is by using the
keypress event for triggering the search view. I hope that in every
case, the keypress event are correctly ordered. It leads to some code
I'm not really proud of this patch, but the key event handling are
quite messy. Also, I need to handle the search string also when the
keyup event is fired, because it might change the search (for example
backspace trigger a keyup, but not a keypress).
The issue is that when a default searchview is requested, it is
initialized without its view_id. Result: debug mode can't edit the
search view. This commit makes sure that when the field_view_get is
received, the correct view_id is set.
Problem was that when the user types quickly in the search bar and press
enter, the keydown event of the enter key happens before the keypress
event of the last key entered. This means that the autocompletion has
a wrong string. The fix is to move the enter selection detection from
keydown to keyup.
search bar does not suggest date field format based on user's locale and always shows based on mmddyy using Date.parse, opw:615276
Note: starting in 9.0, datejs has been replaced by momentjs, so this
problem should be solved in a better way.
The ManyToOneField widget evaluated the domain before doing a name_search,
but the domain should not be evaluated client side, because it can only
be static at that point. It caused crash in autocompletion because
some variables could not be found in the context (see stage_id in
project.task for example)
The domain was using '=', which is correct when selecting a specific
element in autocompletion, but not when searching for all elements
matching a string.
As `_inherits` fields are now handled via `related`
fields (not stored, obviously), a new descriptor
`searchable` has been added to `fields_get()` result
to indicated if the field is searchable or not.
function start, line 783, filters_ready waits for fields_view_get deferred to be resolved before calling prepare_filters, which prepare the view filters.
The thing is that, at line 416, if this.headless boolean is true, the searchview is marked as ready (by resolving the this.ready deferred), and the fields_view_get deferred will never be resolved, as it is resolved in the else case, where the this.headless boolean is false.
Therefore, in this.headless true case, this prevents the deferred $.when(this._super(), filters_ready) to be resolved, as the deferred filters_ready will never be resolved. Therefore, the deferred returned by the start function is never resolved, and the searchview will never be regarded as ready neither.
In views.js, the ViewManager is marked as ready when the deferred manager_ready = $.when(searchview_loaded, main_view_loaded, this.view_completely_inited) is resolved. In the this.headless true case, as the searchview is never marked as ready, the manager isn't neither.
This leaded to some issues, like the action buttons in form views being disabled on click, and never re-enabled once the action/wizard executed. See opw-610723
When going up and wrapping around, the focus was on a hidden separator,
so hitting enter = stacktrace. Now, the focus goes on the last
selectable list item.-
Allow binding an optional `action_id` to filters.
The web client will try to identify the specific
action ID when saving new filters. If no contextual
action exists, the filter is saved globally for
the model.
This will automatically keep filters within their
original menu when there are several menus/actions
leading to a given list of documents.
In some cases the action_id will not match the
filter model, which should be fine (e.g. when opening
a many2one completion popup for model `foo` within
a menu of model `bar`).
It is also still be possible to have a filter apply
to all actions/menus for a given model by manually
deleting the action_id value in the filter
(e.g. via the Manage Filters debug menu).
When updating a filter the action_id value is ignored
so that old global filters will be gradually replaced
by new "local" filters.
Also added an _order to ensure stable ordering of the
filters.
Task #5009
This new autocomplete widget (the one used in the search bar) does not
do remote calls automatically, but on demand. In theory, it should lead
to a better user experience, not having the ui blocked every time long
remote calls are done.
It also has the benefits of bringing us one step closer to not
depending on jquery.ui. Bonus point: the code is quite short (< 200 loc
i believe)
the code handling the keydown events was moved, but the variable this
was not adjusted accordingly, resulting in a broken navigation.
It is now possible to press LEFT and RIGHT again to move the focus
between facets.
the jquery method length() compute the height of its element, even if
it is diplay:none. But in our case, we want the offset to be zero if
there is no visible view_manager_header. (for example, general settings
in settings menu)
Denis Ledoux
Nicolas Lempereur
Olivier Dony
Géry Debongnie
Christophe Simonis
Rifakat Haradwala
Mohammed Shekha
Martin Trigaux