* = hr_timesheet, sale_project, test_main_flows
Smaller changes:
- Projects created on the fly (through `name_create`) will now come with a
default `new` stage in order for them to not be empty.
- Removed task auto assign upon creation besides in FSM's 'My tasks' menu.
- Allow the reordering of projects without needing to group by
anything.
- Make `project.task`.`description` and `project.tags`.`name`
translatable.
- Disable the creation of records in the view when clicking on `Tasks
in recurrence` stat button.
- Track the planned date of the task in the chatter
- Disable the creation of records in the view when clicking on
'invoices' stat button and add the kanban view to that action.
- Make milestones completely available to regular project users.
- Remove the 'Documents' button in the project's kanban settings menu.
- Add kanban, pivot and graph views to the 'Hours Recorded' stat button
on projects
- Add the calendar view on the 'Hours Forecast' stat button on projects
- The 'Sales Orders' stat button on the `project.project`'s form view
will now display the amount of sales order linked to the whole
project. So the one linked to the project itself if it exists + all
the tasks. It will also open them, form view if 1 else list view.
- Fix a typo in the settings 'projets' => 'projects'
- The analytic account of the project will now be assigned to the
sales order when a task is created through the 'Create a task in an
existing project' option.
Changed the portal task view to include a sidebar similar to sales
orders, with a simple menu leading to different parts of the screen.
Changed the `project.task` 'rating' stat button:
- The icon will now represent the latest review.
- The action will directly lead to the record's form view if there is
only 1 rating.
- Make some fields readonly in the form view.
- Display the % of satisfaction instead of the number of ratings.
Reorder all stat buttons on the `project.task` form view in this order:
- Products, Worksheet, Sales Order(s), Invoices, Ratings, Hours
Forecast, Parent Task, Tasks in recurrence, Tickets, Quotations and
lastly Customer Preview
Make the status of the project editable directly through the kanban
view. A new widget has been added to handle that properly. When editing
the status through that means, a `project.update` will be created with
the current date and the appropriate status. In addition to that a new
status has been added (only on `project.task`, not `project.status`)
namely `to_define` in order to differentiate new and running projects.
Projects now start with the `to_define` status.
The `project.project`'s rating stat button has been changed in the
following ways:
- The icon will now change in function of the satisfaction percentage,
smile above 66%, meh between 33% and 66% and frown below 33%.
- The color will also change depending on the rate, smile is green, meh
is orange and frown is red.
- The ratings will now be in function of the last 30 days instead of
all time and the action will also filter on those 30 days.
- Change the action name from 'Rating' to 'Ratings'.
`project.project` ticket stat button:
- Will now open the form view when there is only one record.
- Added the activity view.
- Disable the creation of new records.
- Rename the action to 'Tickets'.
Closes: odoo/odoo#75269
See: odoo/enterprise#20334
Task ID: 2611006
Signed-off-by: Laurent Stukkens (ltu) <ltu@odoo.com>
In https://github.com/odoo/odoo/pull/78652 a fix was made in the layout
designer guaranteeing that the company_details field followed the correct
address format set by the company. However, this fix didn't take into account
all company data, making some `format_address`es raise a `KeyError`. This PR
fixes that by using the company_date defined in `_display_address`.
Fixes https://github.com/odoo/odoo/issues/78942closesodoo/odoo#79097
X-original-commit: 3cdc770c4966cef88362f8119a96de5501d21b18
Signed-off-by: Leonardo Pavan Rocha <lpr@odoo.com>
- The batch version of get_metadata won't work correctly for the xmlid
returns due to the limit=1 (cause by
718b0f8b63 ). Fix the
previous issue by reversing the order and without limit.
- The read should be read with the correct user.
closesodoo/odoo#79091
X-original-commit: 29a53f19a7729f486cbc6ff69893b8d150c3512d
Signed-off-by: Rémy Voet <ryv@odoo.com>
A related field copies the attributes from its target field, except for
the attributes defined on the related field itself. The implementation
of this feature was not working properly for attributes with a truthy
default value.
closesodoo/odoo#79025
X-original-commit: dec4a7ec478fa02f19dee8c8426c88d17dc3c7f3
Signed-off-by: Raphael Collet (rco) <rco@openerp.com>
Issue:
Our functionality for merging PDFs from multiple vendor bills relies on PyPDF2. It is well-known that PyPDF2 is sometimes unable to manipulate even perfectly normal PDFs. To provide a helpful error message to the user when this happens (i.e. give them the names of the vendor bills corresponding to the offending pdfs), the `_get_unreadeable_pdfs()` function is called right before we try to merge the PDFs in `_merge_pdfs`. This function is meant to identify the offending PDFs and provide them to the user.
However, _get_unreadable_pdfs did not notice the problem with 3 of my customer's pdfs, because the error was only triggered once the PdfFileWriter.write function was called in _merge_pdfs. This function, however, is not called in _get_unreadable_pdfs, therefore, it did not notice that anything was wrong.
Fix:
- Change _get_unreadable_pdfs so that it also makes a call to PdfFileWriter.write
- As soon as PdfFileWriter.write fails once, it will continue failing when we append more PDF streams to the PdfFileWriter. I therefore suggest initialising a different PdfFileWriter at each iteration of the for loop in order for an offending PDF to not cause a false positive on subsequent PDFs.
closesodoo/odoo#78966
X-original-commit: 0b4efd37376223806e7ff7273f520b2d17337bc8
Signed-off-by: Nicolas Lempereur (nle) <nle@odoo.com>
Signed-off-by: Antoine Dupuis (andu) <andu@odoo.com>
Some modules are only adding data/tests/routes/static content.
In this case, it is useless to check the database schema or setup models
in registry.
closesodoo/odoo#78898
Signed-off-by: Raphael Collet (rco) <rco@openerp.com>
With a lot of module installed setup_models can be quite slow.
One of the reason for that are the call to resolve_mro.
This operation is fast, but called for a lot of models/fields this
become an important part of get_depends and setup_base.
- This is call once for each models in setup_base
- This is call more than once for each computed_field in get_depends.
The use of a cache shows significative performance improvements.
During the install of a database with all modules, the install times
is reduced by ~3 minutes over ~14
Part-of: odoo/odoo#78898
Co-authored-by: Raphael Collet <rco@odoo.com>
Seems like depending the pillow version, the output image size could differ
from one or two bytes. Now we check the expected size with a tolerance of
+/- 1 byte.
closesodoo/odoo#78965
X-original-commit: 42d3427975592a914b5a61aee304d499750e8afe
Signed-off-by: Jérémy Kersten (jke) <jke@openerp.com>
Some records have NULL create_date. In that case we get a traceback like
below.
```
Traceback (most recent call last):
...
File "/home/odoo/src/odoo/15.0/addons/mail/models/mail_thread.py", line 410, in _compute_field_value
return super()._compute_field_value(field)
File "/home/odoo/src/odoo/15.0/odoo/models.py", line 4249, in _compute_field_value
getattr(self, field.compute)()
File "/home/odoo/src/odoo/15.0/odoo/addons/base/models/res_partner.py", line 259, in _compute_avatar_128
super()._compute_avatar_128()
File "/home/odoo/src/odoo/15.0/odoo/addons/base/models/avatar_mixin.py", line 62, in _compute_avatar_128
self._compute_avatar('avatar_128', 'image_128')
File "/home/odoo/src/odoo/15.0/odoo/addons/base/models/res_partner.py", line 263, in _compute_avatar
super(Partner, partners_with_internal_user)._compute_avatar(avatar_field, image_field)
File "/home/odoo/src/odoo/15.0/odoo/addons/base/models/avatar_mixin.py", line 39, in _compute_avatar
avatar = record._avatar_generate_svg()
File "/home/odoo/src/odoo/15.0/odoo/addons/base/models/avatar_mixin.py", line 66, in _avatar_generate_svg
bgcolor = get_hsl_from_seed(self[self._avatar_name_field] + str(self.create_date.timestamp()))
AttributeError: 'bool' object has no attribute 'timestamp'
```
Observed during upgrade request 40906
closesodoo/odoo#78945
X-original-commit: 559d53fd2a0d437fae595dd4700f2607c8de81a8
Signed-off-by: Sébastien Theys (seb) <seb@odoo.com>
The current system uses an imperative request/response model to a
communication which is full-duplex and thus (hopefully) more suited to
a reactive approach.
This updates the Chrome runner to:
* use a separate thread for reading from Chrome, which limits stalling
/ buffering
* uses futures for request/response cycles
* uses a callbacks system for events (unrequested messages from
chrome)
The test end (success / failure) is also signaled via a future.
This requires a bit more cleanup as the browser is reused for all
tests of a *class*, so e.g. the results future must be reset between
tests.
Also for convenience add debug logging of WS requests / responses,
it's convenient for looking at a test / tour's traffic.
Part-of: odoo/odoo#77735
After some analyze on a lot of customer databases, seems like most of the time
their are performance probleme, and big store, it is due to a lot of big file
uploaded without reason. E.g. barcode, photo, ... a small one will be enough.
Now, we decided (in stable) to auto resize these pictures to 1920x1920px
by default and compress it with a quality of 80 when the source is bigger.
You can bypass this behaviour in your specific use case,
using a context key: 'image_no_postprocess' set to True.
You can disable the resize (and quality implicitely)
using an icp: 'base.image_autoresize_max_px' set to '0'.
You can change the default resize (1920x1920) format using an icp:
'base.image_autoresize_max_px' set to '<width>x<height>' (e.g. '1024x768')
You can change the default quality (80) using an icp:
'base.image_autoresize_quality' with a value between 0 and 100 where 0 skip it.
You can change the type of file that will be post process using icp:
'base.image_autoresize_extensions' (subtype of the mimetype comma separated).
Api of image has not be changed in this commit, only refactored to allow to
work with image directly without the need to encode/Decode in base64 the raw.
We decide to keep 1920x1920 by default instead of 1080p to avoid to resize
portrait picture in 1080px and stay consistent with field image_1920 that
return a 1920px image for width or height whatever the orientation.
+ fix some lint diff for ci style in master
closesodoo/odoo#78556
X-original-commit: d9ce0507960f247e1187baf7bd8399f90be237aa
Signed-off-by: Jérémy Kersten (jke) <jke@openerp.com>
There exists an optimization when creating fields to avoid calling
`setup_models()` for fields when creating a custom model, since the
latter already calls `setup_models()`.
We add the same optimization for field selections, so that creating a
custom field with selections only calls `setup_models()` once. Note
that both optimizations are combined when creating custom models with
selection fields: `setup_models()` will be called once for all.
closesodoo/odoo#78514
Related: odoo/enterprise#21746
Signed-off-by: Xavier Dollé (xdo) <xdo@odoo.com>
Before this commit, trying to create a field related to another field in
the same batch raises an error, because the computation of the field
'related_field_id' cannot find the target field in the registry.
The new approach consists in getting the target field from the database
instead. It works when creating fields in batch, because all records
are inserted into the database before the field is computated.
This costs extra queries, but it allows to batch field creation and
avoids multiple calls to setup_models() when creating a model with
partner field in Studio. (see odoo/enterprise#21746)
The creation of a simple custom related field (related="x.y") costs 3
additional queries to determine the target field:
- 2 queries to get a given field on a given model
- 1 query to get the first field's comodel
Part-of: odoo/odoo#78514
The real overhead of creating a related field is actually two queries:
- one query in update_db_related, the purpose of this test
- one more query for _compute_related_field_id in flush()
With the previous version of this test, one prefetching done in the
first create was not present in the second create since it was in cache.
The fix clears the caches to make the creation of both fields in the
same situation, such that query counts are compared in a fair way.
Part-of: odoo/odoo#78514
Registry updates can be quite expensive (on the order of a second).
When creating new fields, this update is performed *for each field*,
leading to sub-par performances when bulk-creating fields.
Also remove `_existing_field_data` which has been unused since
9afce4805f, and the `clear_caches`
set up for that purpose.
Part-of: odoo/odoo#78514
When using a profiler manually inside the code, and longpolling uses
this part of the code, an error will appear on runbot since gevent
server cannot be profiled properly.
This fix mitigate the issue by disabling the profiler automatically in
this case.
Part-of: odoo/odoo#78514
When installing a database with all modules in enterprise,
install take around 20 minutes and almost half of that is spent in the
`setup_model` method.
There is actually two calls to `setup_models` for each module.
One of them was introduced in b5c50fa824
and only looks useful when upgrading a module with migration scripts.
This first fix proposes to skip `setup_models` if the module state is
`to install`.
closesodoo/odoo#78808
Signed-off-by: Xavier Dollé (xdo) <xdo@odoo.com>
This commit fixes the case where you have some text after a comment.
Until know, we miss it. Now we render the tail part.
```
<t>
<!-- HIDE Text 1 -->
Text 1
<p>SHOW Text 2</p>
</t>
```
After the fix, Text 1 is correctly rendered
This commit fixes#76628closesodoo/odoo#78782
X-original-commit: 566360b07b4ff6c7290f264d9064400df2b06ca5
Signed-off-by: Quentin Smetz (qsm) <qsm@odoo.com>
Signed-off-by: Jérémy Kersten (jke) <jke@openerp.com>
With the release of Debian Bullseye the time has come for the balancing
act by trying to update the requirements.
The constraints are the following:
* Stick as close as possible to python3-* Debian packages versions
of the current Debian stable.
* Same but for the Ubuntu LTS version.
* When one of the above package is patched by Debian or Ubuntu
maintainers, set the upstream version that includes the patch if any.
Also, as support for python < 3.7 is dropped, some cleanup can be done.
The `reportlab / pillow` combo is a special case:
* Pillow has to be updated to 8.1.2 as this version includes the
security patches that were added to Ubuntu package 7.0.0 (Focal).
* Reportlab crashes with 8.1.2 with version prior to 3.5.54 [0].
The problem does not occur on Ubuntu Focal as both versions from
the Ubuntu packaging are compatible.
So the reportlab 3.5.59 is chosen as it's the Debian Bullseye version
and to avoid multiple lines for a few minor versions.
[0] https://hg.reportlab.com/hg-public/reportlab/rev/0cf382dab63b
X-original-commit: 794677fb6a3391379200eb2144a6ed372e89c17a
Part-of: odoo/odoo#78781
This is a followup on b6c688caa2, which
has introduced a new bug.
Consider two groups A and B in a given category, where B implies A and
also A.id > B.id. For the sake of simplicity, assume that A.id=2 and
B.id=1. Following the commit above, the name of the group selection
field will be 'sel_groups_1_2'.
Consider a user in group B. Because B implies A, this means that the
user now belongs to both A and B. A call to read() on that user returns
field 'sel_groups_1_2' with value 2, which on the form view appears as
the user belonging to A only.
The error is in read(), which assumes that the group ids in the field
name correspond to the implication order of the groups, which is no
longer the case since b6c688caa2.
closesodoo/odoo#78662
X-original-commit: e73c406671a51e9c1c4872a9e85eebe5656ea38e
Signed-off-by: Raphael Collet (rco) <rco@openerp.com>
Steps to follow
- Set the thousands_sep of the current lang to an empty string or null
- Create a report
-> The thousand separator from the system locale will be used instead to format numbers
Solution
Use an empty string if there is no thousands separator
opw-2507441
closesodoo/odoo#78384
X-original-commit: 767adf94c3162c1d93d93aa638200f73123f4ab6
Signed-off-by: Nicolas Lempereur (nle) <nle@odoo.com>
Signed-off-by: Hubert Van De Walle <hubvd@users.noreply.github.com>
The processing of the LINK command in a one2many should fail when the
line being linked does not exist. However, there is one case where it
should not fail: when that line existed and was linked before applying
the commands. That use-case may seem strange, but it actually exists:
deleting a line in a sales order automatically deletes the corresponding
reward lines.
closesodoo/odoo#78318
X-original-commit: 4bfe1b5cab10d3171d386d76799a7d7729f49154
Signed-off-by: Raphael Collet (rco) <rco@openerp.com>
Before this commit: the "Asset changed" log message would display
an incorrect version and name. This commit fixes that.
closesodoo/odoo#78175
X-original-commit: d4350ddace643fefa5951655e53f701b13b45333
Signed-off-by: Aaron Bohy (aab) <aab@odoo.com>
As per the logging documentation, it is fine to send arguments that are
not string. Logging takes care of stringifying and injecting the
arguments in the message template.
closesodoo/odoo#78172
Signed-off-by: Xavier Dollé (xdo) <xdo@odoo.com>
base: The two elements were not distinguishable by an xpath.
partner_autocomplete: Use new ids in xpath
The class `o_text_overflow` set on the fields was also
preventing the dropdown menu from appearing. This class is
now moved to the `input` tag instead of the while `div`.
closesodoo/odoo#78141
Task: 2638570
X-original-commit: fd7bfbccd45f874d5a826e695858bc75c8467700
Signed-off-by: Florian Daloze (fda) <fda@odoo.com>
A cursor created by a Connection object is not expected to be used with
environments; use registry.cursor() instead.
closesodoo/odoo#78143
X-original-commit: 6aeb73f10953f3547b7cf830718c02a3933df9e0
Signed-off-by: Raphael Collet (rco) <rco@openerp.com>
In some cases we migh want to change to a value that isn't the default
one when uninstalling.
For instance: when we uninstall the module `event_sale`, a product has
the field `detailed_type` set to `event`, which is a subtype of `service`.
So we want to update the value to `service` when uninstalling instead of
the default value, which would be `consu` and wouldn't make any sense.
Part-of: odoo/odoo#77876
In the payment_adyen module, files of adyen are added as part of the
assets_frontend bundle. That bundle is lazy loaded on the website... but
those external files were not, as not added within the bundle itself.
This commit fixes that, properly lazy loading "remaining" files
out of a lazy loaded assets bundle.
closesodoo/odoo#77877
X-original-commit: 8dd71bdc42c8d4ab3373a2a0601c93d1687aabca
Signed-off-by: Quentin Smetz (qsm) <qsm@odoo.com>
PURPOSE
Help people setuping their mail server with clear labels and form view.
SPECIFICATIONS
Rename Description to Name, as Description indicates a secondary text
field. Add a placeholder to indicate it is used as a functional name
and not a technical field.
Relabel the field for filtering to FROM Filtering. Current From Filter
could lead to think it filters incoming emails which is not the case.
Move button for testing in header as on all form views.
Use radio buttons for authentication and encryption to display available
settings directly to user. This is more user friendly than selection boxes.
Split connection information in two groups: authentication and security.
Each group comes with its options below main radio-based field.
Task-2628092
closesodoo/odoo#76301
Signed-off-by: Thibault Delavallee (tde) <tde@openerp.com>
* [FIX] test_lint: Consider variables for sql-injection
Using the following code:
```python
var = 'SELECT name FROM account WHERE id IN {}'
values = (1, 2, 3)
self._cr.execute(var.format(values))
```
It has a risky sql injection ignored before of this change
And allow psycopg2.SQL way mapping the variables declaration
* [FIX] sql-injection: AttributeError: 'NoneType' object has no attribute 'parent'
Using the following code:
queries = [
"SELECT id FROM res_partner",
"SELECT id FROM res_users",
]
for query in queries:
self.env.cr.execute(query)
The check sql-injection shows the following error:
- AttributeError: 'NoneType' object has no attribute 'parent'
So, Now it is validating if it is not None
* [REF] sql-injection: Using better naming for node_ofc -> node_assign
* [FIX] sql-injection: Fix false positive using BinOp "+"
Considering the following valid case:
cr.execute('SELECT ' + operator + ' FROM table' + 'WHERE')
The representation tree is:
node.repr_tree()
BinOp(
op='+',
left=BinOp(
op='+',
left=BinOp(
op='+',
left=Const(value='SELECT '),
right=Name(name='operator')),
right=Const(value=' FROM table')),
right=Const(value='WHERE'))
Notice that left node is another BinOp node
So, it need to be considered recursively
closesodoo/odoo#77864
X-original-commit: ce1a0171f61d2808470c185a91e9f8299e4efd25
Signed-off-by: Xavier Morel (xmo) <xmo@odoo.com>
The context key 'validation_views' is used by ir.ui.view to determine
what parts of a view arch must be validated. Its associated value is
either True or a recordset, and having those types leads to unexpected
warnings when creating environments.
Before creating a new environment, the class Environment looks for an
existing environment with the given parameters: cr, uid, context, su.
The comparison of a context where 'validation_views' is a recordset with
another context where 'validation_views' is True, generates the warning
"unsupported operand type(s)"...
We avoid this situation by using ids instead of a recordset for the
value of the validation context key.
closesodoo/odoo#77747
X-original-commit: 210cd625a052a4d3c230a93bcba644c9a31e67ab
Signed-off-by: Raphael Collet (rco) <rco@openerp.com>
The exists method (of BaseModel) doesn't work with model
without SQL table but with a table query (see
f2ceef0e2f)
This method is call when we try to read a forbidden
record (`forbidden = missing.exists()` in `_read`)
- Fix it by using a Query object (which able the case).
- Also use a partition method instead of duplicate it in the method.
- Fix the CRM Lead Report which can return a Falsy id
task-2633558
closesodoo/odoo#77644
X-original-commit: 70a68615c722e7145b2ba8bb7ca86a8373b5aa4e
Signed-off-by: Rémy Voet <ryv-odoo@users.noreply.github.com>
Signed-off-by: Raphael Collet (rco) <rco@openerp.com>
This commit removes all the 'extend' initially introduced to avoid code
repetition and ensure visual consistency across Bootstrap and Owl dropdowns.
Despite achieving the desired results, using 'extend' in this context
was seriously impacting the bundle generation time, probably due to an
underestimated amount of Apps' legacy-code applied on these elements.
In order to achieve the same results, the chosen strategy is to add
Bootstrap default classes directly into Owl dropdowns.
Also, it moves code related to bootstrap dropdown in 'webclient.scss',
leaving 'core/dropdown/dropdown.scss' for Owl code only.
Due to the discrepancies between Bootstrap and Owl html
structure, the '.dropdown-item' class could not have been added
directly to Owl's '.o_dropdown_item' itself, without refactoring
the Dropdown component structure.
// ==== Bootstrap 4.6 default Structure ================================
<div class="dropdown-menu">
<button class="dropdown-item" type="button">Action</button>
<a class="dropdown-item" href="#">Another action</a>
</div>
// ==== OWL default Structure before this commit =======================
<ul class="o_dropdown_menu">
<li class="o_dropdown_item">
<span>Action</span>
</li>
<li class="o_dropdown_item">
<a href="#">Another action</a>
</li>
</ul>
// ==== OWL Structure after this commit ================================
<div class="o-dropdown--menu dropdown-menu">
<span class="dropdown-item">Action</span>
<a class="dropdown-item" href="#">Another action</a>
</div>
// ==== web.assets_backend.css Bundle Generation Comparison ============
With all modules installed (enterprise edition over runbot):
Before this commit, bundle took ~2.5s and ~4s to generate and weighted ~322kB (~2.5MB uncompressed)
After this commit, it takes between ~1.2s and ~1.6s and weights ~257kB (~1.6MB uncompressed)
closesodoo/odoo#77649
X-original-commit: 84715436d87bb05b421bc9ccaacda67d07571690
Related: odoo/enterprise#21370
Signed-off-by: Géry Debongnie (ged) <ged@openerp.com>
Co-authored-by: Stefano Rigano <sri@odoo.com>
Co-authored-by: François Georis <fge@odoo.com>
Co-authored-by: Bruno Boi <boi@odoo.com>
Because test cursors are implemented using savepoints, they should *at
most* be nested (ideally they would be strictly sequenced).
If upon being closed a test cursor finds a *different* un-closed
cursor at the top of the stack, one of its followers / children /
descendants was not closed before it, which is a problem.
The initial version of this would look for the cursor being closed in
the stack but this could lead to incoherent cursor stacks and errors
related to the management of the cursors stack, even though it only
exists for reporting reasons.
closesodoo/odoo#76243
Signed-off-by: Xavier Morel (xmo) <xmo@odoo.com>
Initialize the savepoint semi-lazily (on-demand) as cycling a savepoint
generates 4~5 queries (depending whether the savepoint is explicitly
released on COMMIT or not):
SAVEPOINT
-- < do stuff>
-- commit
RELEASE SAVEPOINT -- or not
SAVEPOINT
-- close
ROLLBACK TO SAVEPOINT
RELEASE SAVEPOINT
With a lazy savepoint, this is just 0-1 queries (`SAVEPOINT` at the
first explicit query only, creating a cursor and then closing it
immediately is a no-op).
For reliability use a semi-lazy savepoint: always immediately emit a
`SAVEPOINT` on cursor creation, but don't automatically create one
after each `commit`. That limits the issues of overlapping (but
non-nested) savepoints.
Also fix the `generate` API to not use the request: when `website` was
converted to the new API, `cr`, `uid`, and `context` were dropped as
if it were a model... but it's not. So in order to recover an
execution environment, that was looked up on the session.
That, then, turns out to be an issue when `generate` is triggered from
an RPC call: the RPC layer creates its own cursor and environment
separate from the request's which may not have one at
all. Problematically during testing we're in `mono_db` mode, so the
request's cursor/env can be accessed and will be lazily initialized.
This then causes an issue with the `TestCursor`'s savepoints: rather
than be nested, the lifetimes of the request's and RPC's savepoints
only overlap[0]:
|-- rpc --|
|-- request --|
As a result, when the RPC's cursor is committed and released it
automatically released the request's, and the request's explicit
release then fails. This would break `/website:WithContext.test_search`.
By fixing the API of `generate`, it stops triggering the creation of a
request cursor, and therefore the overlap and resulting error.
[0] the laziness or eagerness of the savepointing in the test cursor
has no impact on this issue, as multiple requests have already been
issued on the RPC's test cursor before the request's is even
created
Part-of: odoo/odoo#76243
The query count increases by one because the old code (with
hand-rolled savepoints) never released the `model_load` savepoint.
Part-of: odoo/odoo#76243
Ensure savepoints are *always* released when exiting the context:
rolling back to a savepoint does not release it, so the savepoint
would remain "active" forever (just possibly shadowed).
Also provide a `Savepoint` object to the user, with the following
facilities:
* `name`, in case there are useful things the user can do with a
savepoint name.
The name uses standard UUID representation (rather
than pure hex) because it's otherwise difficult to differentiate
savepoints: in a UUID1, fields 2, 3, and 5 almost certainly don't
change, and the changes between two UUIDs are the last 2-3 nibbles
of field 1 (`time_low`) and the content of field 4 (`seq`, 4 nibbles
at offset 16), with "properly" separated fields it's much easier to
notice the difference.
* `rollback` allows the user to rollback the savepoint to the
initialisation state at any moment.
* `close` allows the use of `contextlib.closing` as well as closing
the savepoint while in the covered span. Closing a savepoint rolls
it back by default (like cursors).
Because there's no such thing as committing a savepoint, it's also
possible to close *without* rolling back, which is similar (but not
identical).
This requires using the semantics of emitting the `SAVEPOINT` during
object initialisation: `closing` was designed to work with non-CMs so
it does not forward `__enter__` to the wrapped object.
Also introduce a CM type for the flushing:
* `_GeneratorContextManager` does not work well when used outside of a
`with`, especially when it yields something: if the CM itself goes
out of scope, the inner generator is `close`d, which raises a
`GeneratorExit` at the `yield` point, which `__exit__`s whichever CM
is held (and in our case would thus rollback and release the
savepoint)
* we probably want to clear() on `rollback`, so having the flushing CM
extend the savepoint one makes a lot of sense
* while it changes the semantics of `Cursor.savepoint` a bit (now
initialises the savepoint at call time instead of delaying until
`__enter__`), it enables the use of `closing` with flushing, and
without having to use the context manager objects directly:
`Cursor.savepoint()` is the only necessary interface, and should
work fine with and without flushing.
Part-of: odoo/odoo#76243
Enabling sql logging for a specific section of code is currently a
pain in the ass as it requires updating both the cursor and the
logger. This utility does that.
This CM is *not* thread-safe as it updates and resets the cursor
without using CAS or anything, so if cursor 1 gets enabled, then
cursor 2, then cursor 1 stops, then cursor 2, cursor 2 will reset the
logger to `logging.DEBUG` which cursor 1 had set, rather than the
probable `logging.NOTSET` it originally was.
Should be possible to fix by e.g. storing the levels in a static
stack (and setting whatever gets popped) buuut... the logging is
already a bit of a mess when multithreaded so I'm unsure it matters
much.
Part-of: odoo/odoo#76243
Rémy Voet (ryv)