Before this commit, saving a file being watched by watchdog
(--dev=reload) with vim (and possibly other text editors) would
crash the server via a KeyboardInterrupt error.
This happens because vim, emacs cause multiple filesystem events to
happen when saving a single file, the FSWatcher registers these events
as multiple SIGHUPs which in turn calls restart() multiple times in a
very small amount of time which makes the KeyboardInterrupts completely
shut down the server
This is fixed by only calling restart() if odoo.phoenix is False, since
catching a SIGHUP sets it to true.
Fixes#21178
Special thanks to @odony for the help :)
* In Python 3 xlwt apparently does not support writing bytes values ->
try to decode assuming the value may be base64-encoded, this is more
or less the behaviour for CSV exports.
This will most likely not allow the export anyway as Excel cells are
limited to 32k data characters, which accounting for base64
expansion means ~24k worth of data, but that is a pre-existing
issue.
* Also removed support for way outdated browsers from
content_disposition: the Safari case is for Safari 5 (circa 2012)
but versioning apparently changed since then and modern Safari
report their "external" version number rather than the webkit
version number => the current Safari reports version 11, and gets
routed to the "does not support unicode file names", which is
further bugged in Python 3 as it %s's bytes, leading to a resulting
filename of e.g. `b'res.partner.csv'.csv` (with the prefix and
quotes).
* The IE case is for IE8, which has long been unsupported by the web
client.
When you import a record in Odoo, you can specify an id without
prefixing it by the name of the module. e.g. 'project_project_x'
The result of this is an empty string as module name.
(This value is valid for a required field...)
Plus, when you check metadata, you see 'false.project_project_x'.
Now, the default module name is '__import__' instead of an empty string.
We already have similar behavior with '__export__' triggered by
'export_data' method.
With this commit, we change the behaviour of the web client with respect
to spaces in char fields. Most of the time, starting and ending spaces
have no value, and worse, make the data not so reliable.
After this commit, field char will trim by default (so, if the user input a
char as ' abc ', the string 'abc' will be sent to the server instead).
Note that this only applies when the value of the field is changed. If
someone open a form view, then switches to edit mode and save, nothing
will change.
This is the desired behavior most of the time. However, in some rare
cases, this is actually harmful. For example, if we trim the
'decimal_point' field, it will not be possible to enter a whitespace as
decimal separator. In those cases, we introduce a new attribute 'trim',
which allow the developer to desactivate that feature.
Remove weird special case: when the first field of the first line of a one2many
is empty, replace this first field by the comma-separated names of the lines,
and discard the other lines.
Instead of use the original string by default.
Currently odoo export a PO translation file using the logic:
"If there is not a translation then use the source."
This generate a false 100% of file translated for tools based on PO
files.
This commit change this logic to:
"If there is not a translation then use empty string."
Odoo import a PO translation file supporting empty string because
if a item is empty string odoo use the original source.
Closes#17925
With this commit, we add a few extra logs to help debugging issues with
phantomjs.
I suspect that the recent problems with phantomjs are due to the
localstorage not being cleared properly, but the logs are not sufficient
to determine the exact cause.
Before this revision, unlinking a module which has been uninstalled at
least once would raise an Exception because base_module_uninstall
contains a field model_id which is required. The unlink method of
ir_module_module would try to set this field to NULL which is of course
a constraint violation.
This is fixed by manually unlinking the respective base_module_uninstall
records before the unlinking of the ir_module_module record itself
Fixes#20720
DO NOT FORWARD-PORT
Before this commit, whenever an ir_module_module that had been
installed/uninstalled at least once would be unlinked, an Exception
would be raised since the field module_id references an ir_module_module
and it is a required field.
This commit fixes this by adding an ondelete='cascade' to the field.
Fixes#20720
When trying to list db's, the db_user is searched and verified before
searching for the databases.
As it's enclosed in a db.cursor context manager, we can assume that
there is a db user at this very moment.
This commit uses a single query to find databases that belongs to this
cursor user.
Also, this commit adds a message in the logs when an exception occurs
to facilitate debugging instead of silently returning an empty list.
In left menu used for switching between "apps" particular menu, the
apps names were not translated.
This commit use the "string" attribute which is translated instead of
"data-string" fallbacking on data-string (so things still work without
updating the module).
opw-781728
closes#20972
- Install French language
- Set language of Customer A to French
- Create an invoice due in the past
- Open the Customer Statements
The customer statement of A is translated in French, but the amounts are
formatted in the company's user language (English by default).
We add the possiblity to take into account the lang in the context when
calling `formatLang`.
PostgreSQL 10 move sequences' metadata fields into a new `pg_sequence`
system catalog. As a result, selecting from a sequence relation now
only return three fields, and does not include the `increment_by`
metadata field anymore.
This patch obtains the `increment_by` value from the new system catalog
for PostgreSQL server versions >= 10.
Fixes#20269
PostgreSQL 10 move sequences' metadata fields into a new `pg_sequence`
system catalog. As a result, selecting from a sequence relation now
only return three fields, and does not include the `increment_by`
metadata field anymore.
This patch obtains the `increment_by` value from the new system catalog
for PostgreSQL server versions >= 10.
Fixes#20269
get_dsn_parameters may return in some edges cases (like on runbot)
and empty value for the user.
So it should not be used here, instead, a SQL query is able to provide
the currently connected user.
Also, get_dsn_parameters is new in psycopg 2.7.
Thibault Delavallée