It may be confusing that a constraint is not triggered in a name_create call.
Testing the absence of value should be done in an override of create, not in a
api.constrains.
Closes#18382
Before this patch, if some view had both `inherit_id` and `website_id`,
its inheritances were being applied even in different websites.
After this patch, if a view inherits from another, inheritances will only apply
when both views belong to the same (or none) website.
Refactoring based on the pr #17635
This closesodoo/odoo#17635
Without this patch, all Qweb cache is shared among all available websites.
This turns impossible in practice to actually create multiple websites with
different look.
This closesodoo/odoo#18462
In case of foreign language, using unicode characters, the error message may be
badly formated.
e.g.
Unknown error during import: <class 'odoo.exceptions.AccessError'>: เฉพาะผู้ดูแลร..
insted of
Unknown error during import: <class 'odoo.exceptions.AccessError'>: (u'\u0e40\u0e09\u0e1e\u0e32\u0e30\u0e1c\u0e39
opw-766580
Dominican team is focused on Transifex traductions using es.po,
as es_DO.po has a lot of issues in the website and reviewed strings
with wrong wording and mixing english and spanish. From now on
Dominican Republic will inherit es.po and just make specific changes
on github just in the needed words or setences that change base on
region.
Closes#18834
When deciding to prefetch records (getting records from the cache with
no value for the field being fetched), if the field was computed
`determine_value` would just get all records, not limited by the normal
prefetch limit; for large recordsets this would generate gigantic
prefetch lists for records we may not need at all.
Fix by applying the `PREFETCH_MAX` limit to records from the cache as is
done in `_prefetch_field`.
Complementarily, when traversing related fields the prefetch
environment would be lost and every record would get an empty prefetch
environment, so the values would ultimately be read one by one.
Example: select (search) 1000 product.product records, access a
related field (e.g. categ_id) in a loop, on the first iteration the
system would first read 1000 templates, then it would read each
categ_id individually, resulting in >1000 SQL queries rather than the
~2 we would expect.
Fixes#18511
This patch modifies Root#dispatch() in order to keep the httprequest url
in the current thread's attribute and use if for dumpstacks and the
phantomjs tests remaining requests handling.
Also added a counter in order to avoid looping forever when trying to join unclosed http requests
in phantomjs test suite after a phantomjs failure + added a log.warning in such a case so it's easier to troubleshoot runbot's ir.logging.
- Activate multi-company
- Company B is parent of Company A
- Admin is in Company A
- Demo is in Company B
- Create a rate with Demo
The rate is recorded on Company A, not Company B as expected.
This is because we call: `self.env.user._get_company()` which would be
equivalent to `self.env.user.env.user.company_id`. Since the user is
browsed as `sudo`, it returns the company of the Admin.
opw-753139
UserError uses a retrocompatibility function 'except_orm' which one can take a
second parameters. The default value of this second param is None, so if we
don't force an empty string, we will see a None on each raise UserError as last line.
In case Postresgl access rights prevents the pguser to access
`pg_database`, the server should not crash when trying to create the
database because the pguser does not have the right to check if the
database exists and he won't be able to create a new database anyway.
However the database might exists and in such a case the flow should
continue. This patch allows the flow to be resumed in such a case, and
in case the database did not actually exists, a warning will tell the
user that it was not possible to check if the db existed and the
registry for this database will fail to be loaded afterward.
For uploaded modules (e.g. saas), uploaded assets are stored in
attachments.
When the asset content is retrieved, static "filesystem" files are
assumed to be UTF8 and properly decoded, but attachment data is not
and is thus interpreted as ASCII further down the line, leading to a
UnicodeDecodeError (and a broken page/template/...) if the asset data
is not actually ASCII (e.g. if it's UTF-8).
Fix by treating attachment assets as UTF8 encoded, same as filesystem
assets.
OPW-752858
When xlsx is used, the worksheet name may be longer than 31 chars and
raise an exception.
This fix will troncate the name up to 31 chars and will remove invalid
chars.
Similar patch was already done at 0122c05eff for xlwt
xlsxwriter 0.9.8 has added a parameter worksheet_class=None so adding **kw to be
compatible with installs with the latest version (requirements.txt asks for
version 0.9.3)
Closes#18077
opw-751157
When a model contains a field translatable which has a unique
constrain, it is impossible to copy a record. Indeed, the unicity
constrain is always triggered, even if the model takes in charge the
overriding of the copy method to avoid duplication.
Introduced with commit ce7f31b0b0, with the goal to target only
XML/HTML translations. We apply this behavior in case of a callable
translate property, to target only the desired problematic use case.
opw-749481
When importing translations with a csv, the state of the translations was
always set to "To translate" even if the state was "Translated"
due to function _inverse_source.
PS: Making self.source when self is considered as record set could lead to an error.
The return in this function was not needed.
In old API this function was called just for one record.
So this function expect only one record at each call.
opw:747877
It calls `onchange_state()` on `res.partner` which was removed on
(24aba60).
It was added to `res.users` long ago (009ea40) to support a `state_id`
field on the view `res.users.simplified.form` that is no longer there
since v8.0.
The method is then never called and was crashing if called.
Closes#17772
Closes#16973
Steps to reproduce:
1) Go to Settings -> Technical -> User Interface -> Views
2) Create a new view and give only name, leave fields View Type and arch_base empty.
3) Exception occures
Current behaviour:
- The user gets an exception with a full stacktrace.
Expected behaviour:
- The user should get a warning
NUL characters must not be used in query parameters,
as they will be ignored by libpq, being end-of-string
characters.
Preventing NULs avoids unexpected results from
queries. It is only necessary with psycopg2
versions before 2.7, which includes the upstream
fix.
Issue: when using `copy` in a given lang, the value of XML/HTML fields is
copied in its translated form, making existing translations useless. Fix the
value of such fields by storing their original value instead.
The context is not in the kwargs when the integrity error is raised from
a rpc request. So the user lang was lost and the sql error message was not
translated.
The part treating the callable message in sql error is depracated.
opw:743259
Christophe Simonis