The record cache is indexed by `(field, record_id, key)`, where `key` depends
on the environment. The key is either `(cr, uid, context)` or `(cr, uid)`,
depending on whether the field's value is context-dependent. As most fields
are not context-dependent, this should avoid some cache prefetching when
swiching context.
Make `RecordCache` work on a single record only, and implement a mapping from
field names (only) to values.
Also add explicit methods on `RecordCache` to wrap special behavior in cache,
and simplify all special values as a wrapper for a getter function.
The record cache may contain regular and special values. Modify `RecordCache`
so that only methods ending with `_value` check for regular values. The
dictionary methods considers both regular and special values as equivalent.
name in record._cache # test if cache has a value
iter(record._cache) # iterate on values
record._cache[name] # get value
record._cache.get(name) # get value or default
record._cache.has_value(name) # test if cache has a regular value
record._cache.get_value(name) # get regular value or default
1. Create a new company
2. Create a user with access to this company and with access to repair
3. Connect as the new user
4. Try to create a new repair and the stock.warehouse access error
should appear
opw-769160
The server response to an `onchange` on a one2many is a list of
commands. In case of one2many update (command 1), the server
returns all fields (and not only the modified ones).
When saving the record, this is the client responsability to
specify which fields really changed.
Before this commit, all fields were sent at `write` and not only
the changed ones. This was causing issues when the user couldn't
write on the field ; an error was triggered even though the user
didn't change anything.
To fix this, we only tag as changed (in the datapoint `_changes`
attribute) the fields that have a different value after the onchange.
Before this commit, when you clicked on the dropdown, the action
was executed but the dropdown wasn't closed.
This issue was very annoying on mobile because the dropdown is in fullscreen
and we couldn't see the result of the action.
Now, we removed the "stopPropagation" which stops the bootstrap event
to close this dropdown.
for computed street fields.
Qweb reports delete the node if the value is Null but won't if the value is an empty string. And then, l10n_mx_edi module has errors because there are nodes with empty string and the minimum value is one char.
Courtesy of Vauxoo. Was PR #17169
Backport in Saas-14 from opw-770430
Before this commit, the images were messing up the layout of the post, with text uglily lying on their right
This commit solves the issue at rendering of the post, to ensure that current posts on prods be affected in a reversible manner.
Also, we want this commit to have minimal impact
OPW 769721
closes#19322
When the picking type doesn't have use_create_lots or
use_existing_lots the products should end up in one
stock.move.line. The user will then be able to validate the picking
without specifying serial/lot numbers.
sale_payment method to check or create a new transaction when paying
a sale order should be named _check_or_create_sale_tx and kept private
like account_payment one. Both versions exists. This commit fixes
the typo.
Commit 55549f1e61 changed the way bindings between action and
contextual menu entries are generated. However a typo was introduced
linking Invoice Orders to sale order lines instead of sale orders.
This commit improves the use of mail activities by allowing their
integration with the calendar. Activity types now have a category
field that can be used to trigger some specific behavior. First
specific behavior is to have activity types of meeting category.
Creating activities of this type trigger a jump to the calendar to
schedule the meeting. Meeting activities and calender events are
linked to be able to easily navigate through the document, its activities
and its meetings.
Configuration is done using a category on activity types instead of some
hardcoded "meeting" activity to avoid issues with master data and to be
able to choose which activity type should trigger this behavior.
Linking calendar.event to documents is a good idea to allow searching and
filtering meetings. It also allow to have a link to the document from a
meeting and ease navigation.
There are currently some specific links (opportunity_id for crm.lead and
applicant_id for hr.applicant) but this commit allows some generic
document linking through a res_model_id / res_id fields pair.
As unlinking calendar events leads to separate flows depending on the
event being recurrent some fields are recomputed in cache even if records
have already been deleted. This leads to issues when computing recurrence
and accessing some event data. Let us therefore avoid recomputing fields
when unlinking events.
* fix typo, widged is actually widget;
* support name coming from context for calendar event quick create;
Support of context could be made generic in quick create mechanism but
in the meantime this fix is considered as sufficient for the use case
we have.
When unbuilding an order with serial number, the quant with the right
serial number must be found in the system either it raises an UserError
explaining what's the problem.
PS: A problem was found when we tried to unbuild two times the same MO.
After the first unbuild, we deleted it and recreated the same one.
opw:748472
- Set the browser to language 'Spanish (Latin America)' ('es-419')
- As a public user, open a livechat session.
The session doesn't open.
The root cause is the creation of a mail channel which contains
translatable fields (e.g. name). This attempts to create translations
in the 'es_419' language. However, this locale doesn't exist in Odoo,
and it triggers the constraint 'lang_fkey_res_lang'.
In this specific case, there is no real reason to translate the channel
name. Therefore, we simply avoid it thanks to a context key.
opw-766489
- datepicker doesn't default to current date by default
- date string in the input is now selected when clicking the field
- adapted the existing tests to the new changes
When a x2m field is modified, the 'save' rpc doesn't send 'False' as second
argument of the special command that manipulate the set of records.
From now on, the virtual id of the record is sent to the server as second
argument. This commit fixes the tests making the assumption that 'False'
was sent.
Fields sent from the client are pretty much always *text*. This can be
problematic for image fields sent as base64 *text* as in Python 3 the
base64 routines only work between bytes.
Alter the core field-resizing function (image_get_resized_images) such
that it ensures text base64 sources are encoded back to bytes before
being forwarded to the main resizing routines (which I don't think
should deal with *that* issue).
Fixes#18985: error when trying to update a company's logo
There is some odd behavior especially inside a editable containing
newlines.
For example in the sale description on an ecommerce item, lines could
contain newline (it is editable from backend) but as an effect of other
use case, pressing enter could have a range of unwanted side effects:
- removing all the content,
- duplicating the editable (should never happen, if we have one field
"description_sale" we don't want to duplicate it, this has no sense)
In a product name which is in the same use case, there is no newline so
the problem is not apparent, and ENTER do nothing. This is the behavior
we should want in every instance.
If we still want a newline, we have to force it with SHIFT + ENTER key,
also this is regarding field that may or could be present in the backend
views, so it could also be edited in backend.
This change make ENTER ignored when we are editing a node directly child
of an node being edited.
A second part of the issue was that dom.removeBetween (used to remove
content inside a range) could remove the ancestor node that was an
editable. This is not wanted.
For example if we did CTRL + A inside a field with `<br />` then ENTER
or pasting something, the field would be removed and not just emptied
and could not be gotten back without editing the template manually.
This was sometimes mitigated because if a `<br />` was at the end of an
element, it would stay in this use case avoiding the issue.
note: not necessary after saas-16
fixes#17445
opw-751360
closes#19313
qdp-odoo