PURPOSE
This commit removes the mass_mailing.campaign model. Instead of having a fully
fledged model, we will simply inherit utm.campaign. We will also add relevant
statistics on utm campaign model in order to use it in various applications.
SPECIFICATIONS
This commit adds leads/generated statistics in the kanban and
the form view of the utm campaign.
Kanban and form views will now display the number of leads and
opportunities generated with the campaign
TaskID: 2002029
PR: #34452
PURPOSE
This commit removes the mass_mailing.campaign model. Instead of having a fully
fledged model, we will simply inherit utm.campaign. We will also add relevant
statistics on utm campaign model in order to use it in various applications.
SPECIFICATIONS
This commit removes the mass_mailing.campaign model. Instead of having a fully
fledged model, we will simply inherit utm.campaign. This change implies that
mass_mailing.tag and mass_mailing.stage have to move to the utm model along
their associated views/data.
These changes were made so that campaigns could be used in the future
by social, mass_mailing and mass_sms and available in the same view
This commit also removes the source_id and the medium_id
fields on the campaign.
This commit also moves the unique_ab_testing field from the mass_mailing_campaign
to the mass_mailing model
Task ID: 2002029
PR: #34015
Task 2038326
* Tick the credit note sequence by default on the customer invoices and vendor bills journals
* Write Customer Invoice/Vendor Bill/Credit Note/Customer refund at the top of the document
* The reference field should be printed on the credit note pdf (interesting because it gives source and reason)
* If the document is a credit note, there shouldn't be any payment information on the pdf (the customer doesn't have to pay you) (currently you see the "payment communication" the customer is supposed to use)
* Rename Credit Note to Refund on dashboard
* Add tooltip on payment widget
closesodoo/odoo#35231
Signed-off-by: Cedric Snauwaert (csn) <csn@openerp.com>
If you have a PST number, certain taxes
don't need to be paid, but the number
should be indicated on the invoice however.
opw-1951352
Closes#31982
Signed-off-by: Cedric Snauwaert (csn) <csn@openerp.com>
If the connection test to mail server succeeds, we shouldn't see
"Oops something went wrong" and "User Error" either
Replace error generation to display success message by bus notify.
closesodoo/odoo#35292
Signed-off-by: Yannick Tivisse (yti) <yti@odoo.com>
When a user change his langage in employee profile, he doesn't see
the result before reloading manually the page.
With this commit, when a user change his language, the page is
reloaded on saving. And remove Create button on My Profile
closesodoo/odoo#34906
Signed-off-by: Yannick Tivisse (yti) <yti@odoo.com>
* account, hr_contract, hr_expense, point_of_sale, sale_margin,
website_sale_delivery, website_sale_wishlist
digits only works on Float fields, on Monetary fields it has no
effect:
- The column_type of a monetary field is always numeric
- It is not applied in `convert_to_column()` or `convert_to_cache()`
- It is not part of `description_attrs`, so it will not be included in
the output of `fields_get()`
closesodoo/odoo#35336
Signed-off-by: Martin Trigaux (mat) <mat@odoo.com>
We added a button for the front end closure to go back to the back end.
Before this, the only way to go back was to change the URL, now we can close the session with the frontend properly.
Also moved the css related to pos_hr in the good directory.
Task-id: 2032178
closesodoo/odoo#34948
Signed-off-by: pimodoo <pimodoo@users.noreply.github.com>
When we try to validate a pos session without beinig administrator, we
don't have access rights on the valuation layer.
So we've extended this rev: 7ef11e3c96
Before this commit, when clicking on a cell in the listview, the cell
receives the focus, which makes it have an extra style (darker gray)
when waiting for the page to laod.
After this commit, the extra style will only be added when using the
keyboard to navigate between the cells and not on click.
task-id: 2041902
closesodoo/odoo#35303
Signed-off-by: Mathieu Duckerts-Antoine <Polymorphe57@users.noreply.github.com>
This appears to be leftover from a C/P from the line above, however
`type="html"` does not make much sense on `<p>` element.
closesodoo/odoo#35124
Signed-off-by: Martin Trigaux (mat) <mat@odoo.com>
* account, crm, hr, im_livechat, l10n_ch, l10n_it_edi, mail,
mass_mailing, project, stock, base
- `String` -> `string`;
- `defaut` -> `default`;
- `defaults` -> `default`;
- `reandonly` -> `readonly`;
- removed redundant `placeholder` attribute for `Char` field;
- removed redundant `size` attribute for `Float` and `Integer` fields;
- changed to use `Selection` instead of `Char` for a field having a
defined `selection`.
closesodoo/odoo#35356
Signed-off-by: Martin Trigaux (mat) <mat@odoo.com>
The group 'group_discount_per_so_line' is used on reports in
point_of_sale but is created in 'sale' module and point_of_sale doesn't
depends on sale. So we've move this group on product to be able to
properly use it in point_of_sale.
TASK-ID: 2008468
closesodoo/odoo#35041
Signed-off-by: pimodoo <pimodoo@users.noreply.github.com>
The stat button on the pos session that shows the different POS orders
linked to the session has been deleted by mistake in rev: 91532b7b13
So we've set it back.
We made here some small changes on the views of different object of POS
module, and set the field journal_ids required in the views, because
launching a POS without payment method won't magically assign one
anymore.
TASK-ID: 2008468
When the accounting localisation is installed after the POS config, the
values on the default POS config which is in data are not completed
because the POS config has been created before any journal exists.
So now, when a chart template is installed, we are completing the POS
config values, and creating the Sale POS journal. We also call the
creation of POS journal when the point_of_sale module is installed after
the template is installed on the database to be sure that we have a
default value for POS journal.
TASK-ID: 2008468
When a pos config is created, default values set on it are not allways
correct, for the sale journal we are referencing a data, which mean
rthat it only worls in the first company. For the invoice journal, we
are referencing the wrong journal in the onchannge, etc
TASK-ID: 2008468
When a chart of account is installed, it first remove all existing
existing data. If you don't have account_accountant module installed,
but you have create a bank or cash journal (through the POS for
example). Some accounts have been created and the installation of chart
template is trying to remove some models, but as the account_accountant
you don't have access write on model 'account.reconcile.model'.
No need for selection fields
If in global variable, the _lt should be used instead
closesodoo/odoo#31211
Signed-off-by: Martin Trigaux (mat) <mat@odoo.com>
Introduces the method _lt
The translation method is now evaluated lazily.
It allows to declare global variables with translatable content
e.g. this code will now work:
LABEL = _lt("User")
def _compute_label(self):
context = {'lang': self.partner_id.lang}
self.user_label = LABEL
Since the introduction of the stock valuation layer, unit_cost is
rounded in the company currency (actually it is NOT because of an ORM
bug that will be fixed later). With the test introduced at [0], fixing
that ORM bug shows we have a rounding issue (27.02 instead of 27) due to
the company currency rounding on the unit_cost. Since the logic of the
anglo saxon will probably always give some rounding issue (we always
return a price unit that will be multiply by the invoiced quantity)
recomputing a not rounded price unit by using the not rounded value with
the not rounded quantity seems a good enough compromise.
[0] 13a535a091closesodoo/odoo#35384
Signed-off-by: Simon Lejeune (sle) <sle@openerp.com>
In this commit -
1. Previously, date_planned of PO was computed and set while the creation of PO.
In this commit, removed the compute logic and now field will be empty until
user fills it.
2. ‘Set date to all order lines’ button is removed.
3. Now if user have filled the date_planned of PO, it will be copied to all po
lines.
4. Schedule date of PO line will be readonly if date_planned have any value
else will be editable.
5. Update the test cases of computation of date_planned.
Task-2032417
This commit adds new `embed_viewer` widget which will be used display URLs
via embedded iframes. You can add any valid URL to display page via iframe,
although widget automatically converts raw google slide URL into embedded URL.
Task ID: 1909010
closesodoo/odoo#29404
Signed-off-by: Simon Lejeune (sle) <sle@openerp.com>
Co-authored-by: pga-odoo <pga@odoo.com>
Those two decorators are removed/deprecated since recent commits but
some references remained in the documentation.
api.guess and api.noguess is deprecated and removed since c552fb7a61closesodoo/odoo#35332
Signed-off-by: Martin Trigaux (mat) <mat@odoo.com>
- Update the CoA for 2020
- Add the most recent monthly tax report
instead of the old one with the new tax model
- Base language English (multilang translation in fr and de)
hsh and rgo co-authored
opw-1917932
closesodoo/odoo#35371
Signed-off-by: Josse Colpaert <jco@openerp.com>
1/ Improve res.partner search for customers/suppliers
=========================================
Purpose
-------
This commit allows to call `name_search` with context key
`res_partner_search_mode` that can take currently two values: "customer" or
"supplier".
When ordering search results, order partners by the number of SO they have when
the value is "customer" and by the number of PO if the value is "supplier".
Top suppliers/customers are displayed above in a PO/SO partner dropdown.
We have thought about different implementations before selecting this one:
- Keep the 2 boolean flags and automatically set them if a SO or PO is created
for a customer/supplier (helps a bit but doesn't work well for the first
SO/PO). As the onboarding is also a priority, we didn't implement it.
- Instead of filtering on the bool flags, sort the name_search() result according
to the number of orders they already have (e.g. if you search for "Foo" on a
PO, a supplier "SuperFoo" with 32 purchase orders will appear before supplier
"HorribleFoo" that has no PO). Problem is to do this in an efficient manner for
large databases - possibly we could store a "relevance index" based on the
number
of recent orders. Storing those values is an issue too, as it doesn't work well
for multi company database. Indeed, storing values that depend on the context
is certainly not a good idea.
- Same as previous alternative but with no stored columns, rather a JOIN in
name_search. This is the solution that have been kept. The only concern was
about the perfomances. We tested it on our production base, on a sales order,
as we have way much more customers than suppliers. After a deep analysis on
the querry and the bitmap generated by PostgreSQL, we can conclude that the
count(*) works rather efficiently and that the request is not significantly
heavier than without the context key.
Performance Analysis
--------------------
The following results have been obtained on our production database.
### Before this commit ###
The original query is the following one:
```
EXPLAIN ANALYSE
SELECT res_partner.id
FROM "res_partner"
WHERE ("res_partner"."customer" = True) AND ("res_partner"."active" = True) AND ((("res_partner"."type" != 'private') OR "res_partner"."type" IS NULL) OR "res_partner"."type" IS NULL ) AND (res_partner.email ilike '%eezee-i%'
OR res_partner.display_name ilike '%eezee-i%'
OR res_partner.ref ilike '%eezee-i%'
OR res_partner.vat ilike '%eezeei%')
-- don't panic, trust postgres bitmap
ORDER BY res_partner.display_name ilike '%eezee-i%' desc,
res_partner.display_name
limit 8;
```
Here is the EXPLAIN ANALYSE returned by PostgreSQL:
```
Limit (cost=1625.50..1625.52 rows=8 width=21) (actual time=13.397..13.404 rows=8 loops=1)
-> Sort (cost=1625.50..1626.32 rows=326 width=21) (actual time=13.395..13.396 rows=8 loops=1)
Sort Key: (((display_name)::text ~~* '%eezee-i%'::text)) DESC, display_name
Sort Method: top-N heapsort Memory: 26kB
-> Bitmap Heap Scan on res_partner (cost=1282.79..1618.98 rows=326 width=21) (actual time=13.025..13.350 rows=56 loops=1)
Recheck Cond: (((email)::text ~~* '%eezee-i%'::text) OR ((display_name)::text ~~* '%eezee-i%'::text) OR ((ref)::text ~~* '%eezee-i%'::text) OR ((vat)::text ~~* '%eezeei%'::text))
Rows Removed by Index Recheck: 1
Filter: (customer AND active AND (((type)::text <> 'private'::text) OR (type IS NULL) OR (type IS NULL)))
Rows Removed by Filter: 1
Heap Blocks: exact=58
-> BitmapOr (cost=1282.79..1282.79 rows=328 width=0) (actual time=12.983..12.983 rows=0 loops=1)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..322.22 rows=162 width=0) (actual time=5.022..5.022 rows=54 loops=1)
Index Cond: ((email)::text ~~* '%eezee-i%'::text)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..322.22 rows=162 width=0) (actual time=4.128..4.128 rows=56 loops=1)
Index Cond: ((display_name)::text ~~* '%eezee-i%'::text)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..321.00 rows=1 width=0) (actual time=2.240..2.240 rows=0 loops=1)
Index Cond: ((ref)::text ~~* '%eezee-i%'::text)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..317.03 rows=4 width=0) (actual time=1.585..1.585 rows=0 loops=1)
Index Cond: ((vat)::text ~~* '%eezeei%'::text)
Planning time: 0.766 ms
Execution time: 13.489 ms
```
### After this commit ###
Without the context key, the query looks like this:
```
SELECT res_partner.id
FROM "res_partner"
WHERE ("res_partner"."active" = True) AND ((("res_partner"."type" != 'private') OR "res_partner"."type" IS NULL) OR "res_partner"."type" IS NULL ) AND (res_partner.email ilike '%eezee-i%'
OR res_partner.display_name ilike '%eezee-i%'
OR res_partner.ref ilike '%eezee-i%'
OR res_partner.vat ilike '%eezeei%')
-- don't panic, trust postgres bitmap
GROUP BY res_partner.id
ORDER BY COUNT(*) DESC, res_partner.display_name ilike '%eezee-i%' desc,
res_partner.display_name
limit 8;
```
And it quite clear when looking at the EXPLAIN ANALYSE that the request
is quite the same, except the aggregation that is made with a quicksort method.
```
Limit (cost=1645.00..1645.02 rows=8 width=29) (actual time=13.200..13.206 rows=8 loops=1)
-> Sort (cost=1645.00..1645.82 rows=328 width=29) (actual time=13.197..13.198 rows=8 loops=1)
Sort Key: (count(*)) DESC, (((display_name)::text ~~* '%eezee-i%'::text)) DESC, display_name
Sort Method: top-N heapsort Memory: 26kB
-> GroupAggregate (cost=1631.88..1638.44 rows=328 width=29) (actual time=13.090..13.160 rows=56 loops=1)
Group Key: id
-> Sort (cost=1631.88..1632.70 rows=328 width=20) (actual time=13.080..13.084 rows=56 loops=1)
Sort Key: id
Sort Method: quicksort Memory: 29kB
-> Bitmap Heap Scan on res_partner (cost=1282.79..1618.17 rows=328 width=20) (actual time=12.793..13.055 rows=56 loops=1)
Recheck Cond: (((email)::text ~~* '%eezee-i%'::text) OR ((display_name)::text ~~* '%eezee-i%'::text) OR ((ref)::text ~~* '%eezee-i%'::text) OR ((vat)::text ~~* '%eezeei%'::text))
Rows Removed by Index Recheck: 1
Filter: (active AND (((type)::text <> 'private'::text) OR (type IS NULL) OR (type IS NULL)))
Rows Removed by Filter: 1
Heap Blocks: exact=58
-> BitmapOr (cost=1282.79..1282.79 rows=328 width=0) (actual time=12.755..12.755 rows=0 loops=1)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..322.22 rows=162 width=0) (actual time=4.823..4.823 rows=54 loops=1)
Index Cond: ((email)::text ~~* '%eezee-i%'::text)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..322.22 rows=162 width=0) (actual time=3.941..3.941 rows=56 loops=1)
Index Cond: ((display_name)::text ~~* '%eezee-i%'::text)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..321.00 rows=1 width=0) (actual time=2.186..2.186 rows=0 loops=1)
Index Cond: ((ref)::text ~~* '%eezee-i%'::text)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..317.03 rows=4 width=0) (actual time=1.798..1.798 rows=0 loops=1)
Index Cond: ((vat)::text ~~* '%eezeei%'::text)
Planning time: 0.761 ms
Execution time: 13.317 ms
```
With the context key, the query looks like this:
```
EXPLAIN ANALYSE
SELECT res_partner.id
FROM "res_partner"
LEFT JOIN sale_order ON res_partner.id = sale_order.partner_id
WHERE ("res_partner"."active" = True) AND ((("res_partner"."type" != 'private') OR "res_partner"."type" IS NULL) OR "res_partner"."type" IS NULL ) AND (res_partner.email ilike '%eezee-i%'
OR res_partner.display_name ilike '%eezee-i%'
OR res_partner.ref ilike '%eezee-i%'
OR res_partner.vat ilike '%eezeei%')
-- don't panic, trust postgres bitmap
GROUP BY res_partner.id
ORDER BY COUNT(*) DESC, res_partner.display_name ilike '%eezee-i%' desc,
res_partner.display_name
limit 8;
```
The only difference is that a nested loop is made for the left join, as postgreSQL has
correctly identified the dicriminating table. The correct result is obtained within
a similar duration.
```
Limit (cost=2700.68..2700.70 rows=8 width=29) (actual time=12.561..12.568 rows=8 loops=1)
-> Sort (cost=2700.68..2701.50 rows=328 width=29) (actual time=12.559..12.560 rows=8 loops=1)
Sort Key: (count(*)) DESC, (((res_partner.display_name)::text ~~* '%eezee-i%'::text)) DESC, res_partner.display_name
Sort Method: top-N heapsort Memory: 26kB
-> GroupAggregate (cost=2687.56..2694.12 rows=328 width=29) (actual time=12.449..12.527 rows=56 loops=1)
Group Key: res_partner.id
-> Sort (cost=2687.56..2688.38 rows=328 width=20) (actual time=12.408..12.425 rows=305 loops=1)
Sort Key: res_partner.id
Sort Method: quicksort Memory: 42kB
-> Nested Loop Left Join (cost=1283.21..2673.86 rows=328 width=20) (actual time=11.629..12.340 rows=305 loops=1)
-> Bitmap Heap Scan on res_partner (cost=1282.79..1618.17 rows=328 width=20) (actual time=11.596..11.862 rows=56 loops=1)
Recheck Cond: (((email)::text ~~* '%eezee-i%'::text) OR ((display_name)::text ~~* '%eezee-i%'::text) OR ((ref)::text ~~* '%eezee-i%'::text) OR ((vat)::text ~~* '%eezeei%'::text))
Rows Removed by Index Recheck: 1
Filter: (active AND (((type)::text <> 'private'::text) OR (type IS NULL) OR (type IS NULL)))
Rows Removed by Filter: 1
Heap Blocks: exact=58
-> BitmapOr (cost=1282.79..1282.79 rows=328 width=0) (actual time=11.560..11.560 rows=0 loops=1)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..322.22 rows=162 width=0) (actual time=4.724..4.724 rows=54 loops=1)
Index Cond: ((email)::text ~~* '%eezee-i%'::text)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..322.22 rows=162 width=0) (actual time=3.825..3.825 rows=56 loops=1)
Index Cond: ((display_name)::text ~~* '%eezee-i%'::text)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..321.00 rows=1 width=0) (actual time=1.752..1.752 rows=0 loops=1)
Index Cond: ((ref)::text ~~* '%eezee-i%'::text)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..317.03 rows=4 width=0) (actual time=1.254..1.254 rows=0 loops=1)
Index Cond: ((vat)::text ~~* '%eezeei%'::text)
-> Index Only Scan using sale_order_partner_id_index on sale_order (cost=0.42..2.74 rows=48 width=4) (actual time=0.006..0.007 rows=5 loops=56)
Index Cond: (partner_id = res_partner.id)
Heap Fetches: 12
Planning time: 1.056 ms
Execution time: 13.791 ms
```
2/ Remove customer and supplier fields from res.partner
===========================================
Purpose
-------
Fields `customer` and `supplier` on `res.partner`
are mostly used in domains of many2x fields.
Those domains can confuse end users because they don't
see the partner they are looking for; and it's not obvious why.
Some identified problems:
1. It can lead to duplicated partners: the user does not find
the partner, so he creates a new one.
2. The user imports supplier contacts in the Contacts app, so they
don't get the `supplier` flag. Then the user wants to make a purchase order,
and cannot find the new suppliers in the list
3. A user removes the customer flag on a prospect, because they don't think
it's a customer yet - except now they can't make a quote for that customer...
Specification
-------------
Remove the two mentioned fields.
Since fields `customer` and `supplier` have been removed, all partners
are now shown in many2one dropdowns.
But in some cases, not all partners are relevant or some are more likely
to be relevant than others. e.g. when creating a PO, top suppliers have a
higher priority than other partners.
So, adapt the places where those fields were used with the new mechanism to
display the searched the partners, according to the number purchase/sales
orders they made.
TaskID: 2031147
closesodoo/odoo#34524
Signed-off-by: Yannick Tivisse (yti) <yti@odoo.com>
Co-authored-by: Yannick Tivisse <yti@odoo.com>
Purpose
=======
Fields `customer` and `supplier` on `res.partner`
are mostly used in domains of many2x fields.
Those domains can confuse end users because they don't
see the partner they are looking for; and it's not obvious why.
Some identified problems:
1. It can lead to duplicated partners: the user does not find
the partner, so he creates a new one.
2. The user imports supplier contacts in the Contacts app, so they
don't get the `supplier` flag. Then the user wants to make a purchase order,
and cannot find the new suppliers in the list
3. A user removes the customer flag on a prospect, because they don't think
it's a customer yet - except now they can't make a quote for that customer...
Specification
=============
Remove the two mentioned fields.
Since fields `customer` and `supplier` have been removed, all partners
are now shown in many2one dropdowns.
But in some cases, not all partners are relevant or some are more likely
to be relevant than others. e.g. when creating a PO, top suppliers have a
higher priority than other partners.
So, adapt the places where those fields were used with the new mechanism to
display the searched the partners, according to the number purchase/sales
orders they made.
TaskID: 2031147
Co-authored-by: Yannick Tivisse <yti@odoo.com>
This commit allows to call `name_search` with context key
`res_partner_search_mode` that can take currently two values: "customer" or
"supplier".
When ordering search results, order partners by the number of SO they have when
the value is "customer" and by the number of PO if the value is "supplier".
Top suppliers/customers are displayed above in a PO/SO partner dropdown.
We have thought about different implementations before selecting this one:
- Keep the 2 boolean flags and automatically set them if a SO or PO is created
for a customer/supplier (helps a bit but doesn't work well for the first
SO/PO). As the onboarding is also a priority, we didn't implement it.
- Instead of filtering on the bool flags, sort the name_search() result according
to the number of orders they already have (e.g. if you search for "Foo" on a
PO, a supplier "SuperFoo" with 32 purchase orders will appear before supplier
"HorribleFoo" that has no PO). Problem is to do this in an efficient manner for
large databases - possibly we could store a "relevance index" based on the
number
of recent orders. Storing those values is an issue too, as it doesn't work well
for multi company database. Indeed, storing values that depend on the context
is certainly not a good idea.
- Same as previous alternative but with no stored columns, rather a JOIN in
name_search. This is the solution that have been kept. The only concern was
about the perfomances. We tested it on our production base, on a sales order,
as we have way much more customers than suppliers. After a deep analysis on
the querry and the bitmap generated by PostgreSQL, we can conclude that the
count(*) works rather efficiently and that the request is not significantly
heavier than without the context key.
Performance Analysis
====================
The following results have been obtained on our production database.
Before this commit
------------------
The original query is the following one:
```
EXPLAIN ANALYSE
SELECT res_partner.id
FROM "res_partner"
WHERE ("res_partner"."customer" = True) AND ("res_partner"."active" = True) AND ((("res_partner"."type" != 'private') OR "res_partner"."type" IS NULL) OR "res_partner"."type" IS NULL ) AND (res_partner.email ilike '%eezee-i%'
OR res_partner.display_name ilike '%eezee-i%'
OR res_partner.ref ilike '%eezee-i%'
OR res_partner.vat ilike '%eezeei%')
-- don't panic, trust postgres bitmap
ORDER BY res_partner.display_name ilike '%eezee-i%' desc,
res_partner.display_name
limit 8;
```
Here is the EXPLAIN ANALYSE returned by PostgreSQL:
```
Limit (cost=1625.50..1625.52 rows=8 width=21) (actual time=13.397..13.404 rows=8 loops=1)
-> Sort (cost=1625.50..1626.32 rows=326 width=21) (actual time=13.395..13.396 rows=8 loops=1)
Sort Key: (((display_name)::text ~~* '%eezee-i%'::text)) DESC, display_name
Sort Method: top-N heapsort Memory: 26kB
-> Bitmap Heap Scan on res_partner (cost=1282.79..1618.98 rows=326 width=21) (actual time=13.025..13.350 rows=56 loops=1)
Recheck Cond: (((email)::text ~~* '%eezee-i%'::text) OR ((display_name)::text ~~* '%eezee-i%'::text) OR ((ref)::text ~~* '%eezee-i%'::text) OR ((vat)::text ~~* '%eezeei%'::text))
Rows Removed by Index Recheck: 1
Filter: (customer AND active AND (((type)::text <> 'private'::text) OR (type IS NULL) OR (type IS NULL)))
Rows Removed by Filter: 1
Heap Blocks: exact=58
-> BitmapOr (cost=1282.79..1282.79 rows=328 width=0) (actual time=12.983..12.983 rows=0 loops=1)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..322.22 rows=162 width=0) (actual time=5.022..5.022 rows=54 loops=1)
Index Cond: ((email)::text ~~* '%eezee-i%'::text)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..322.22 rows=162 width=0) (actual time=4.128..4.128 rows=56 loops=1)
Index Cond: ((display_name)::text ~~* '%eezee-i%'::text)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..321.00 rows=1 width=0) (actual time=2.240..2.240 rows=0 loops=1)
Index Cond: ((ref)::text ~~* '%eezee-i%'::text)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..317.03 rows=4 width=0) (actual time=1.585..1.585 rows=0 loops=1)
Index Cond: ((vat)::text ~~* '%eezeei%'::text)
Planning time: 0.766 ms
Execution time: 13.489 ms
```
After this commit
-----------------
Without the context key, the query looks like this:
```
SELECT res_partner.id
FROM "res_partner"
WHERE ("res_partner"."active" = True) AND ((("res_partner"."type" != 'private') OR "res_partner"."type" IS NULL) OR "res_partner"."type" IS NULL ) AND (res_partner.email ilike '%eezee-i%'
OR res_partner.display_name ilike '%eezee-i%'
OR res_partner.ref ilike '%eezee-i%'
OR res_partner.vat ilike '%eezeei%')
-- don't panic, trust postgres bitmap
GROUP BY res_partner.id
ORDER BY COUNT(*) DESC, res_partner.display_name ilike '%eezee-i%' desc,
res_partner.display_name
limit 8;
And it quite clear when looking at the EXPLAIN ANALYSE that the request
is quite the same, except the aggregation that is made with a quicksort method.
Limit (cost=1645.00..1645.02 rows=8 width=29) (actual time=13.200..13.206 rows=8 loops=1)
-> Sort (cost=1645.00..1645.82 rows=328 width=29) (actual time=13.197..13.198 rows=8 loops=1)
Sort Key: (count(*)) DESC, (((display_name)::text ~~* '%eezee-i%'::text)) DESC, display_name
Sort Method: top-N heapsort Memory: 26kB
-> GroupAggregate (cost=1631.88..1638.44 rows=328 width=29) (actual time=13.090..13.160 rows=56 loops=1)
Group Key: id
-> Sort (cost=1631.88..1632.70 rows=328 width=20) (actual time=13.080..13.084 rows=56 loops=1)
Sort Key: id
Sort Method: quicksort Memory: 29kB
-> Bitmap Heap Scan on res_partner (cost=1282.79..1618.17 rows=328 width=20) (actual time=12.793..13.055 rows=56 loops=1)
Recheck Cond: (((email)::text ~~* '%eezee-i%'::text) OR ((display_name)::text ~~* '%eezee-i%'::text) OR ((ref)::text ~~* '%eezee-i%'::text) OR ((vat)::text ~~* '%eezeei%'::text))
Rows Removed by Index Recheck: 1
Filter: (active AND (((type)::text <> 'private'::text) OR (type IS NULL) OR (type IS NULL)))
Rows Removed by Filter: 1
Heap Blocks: exact=58
-> BitmapOr (cost=1282.79..1282.79 rows=328 width=0) (actual time=12.755..12.755 rows=0 loops=1)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..322.22 rows=162 width=0) (actual time=4.823..4.823 rows=54 loops=1)
Index Cond: ((email)::text ~~* '%eezee-i%'::text)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..322.22 rows=162 width=0) (actual time=3.941..3.941 rows=56 loops=1)
Index Cond: ((display_name)::text ~~* '%eezee-i%'::text)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..321.00 rows=1 width=0) (actual time=2.186..2.186 rows=0 loops=1)
Index Cond: ((ref)::text ~~* '%eezee-i%'::text)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..317.03 rows=4 width=0) (actual time=1.798..1.798 rows=0 loops=1)
Index Cond: ((vat)::text ~~* '%eezeei%'::text)
Planning time: 0.761 ms
Execution time: 13.317 ms
```
With the context key, the query looks like this:
```
EXPLAIN ANALYSE
SELECT res_partner.id
FROM "res_partner"
LEFT JOIN sale_order ON res_partner.id = sale_order.partner_id
WHERE ("res_partner"."active" = True) AND ((("res_partner"."type" != 'private') OR "res_partner"."type" IS NULL) OR "res_partner"."type" IS NULL ) AND (res_partner.email ilike '%eezee-i%'
OR res_partner.display_name ilike '%eezee-i%'
OR res_partner.ref ilike '%eezee-i%'
OR res_partner.vat ilike '%eezeei%')
-- don't panic, trust postgres bitmap
GROUP BY res_partner.id
ORDER BY COUNT(*) DESC, res_partner.display_name ilike '%eezee-i%' desc,
res_partner.display_name
limit 8;
```
The only difference is that a nested loop is made for the left join, as postgreSQL has
correctly identified the dicriminating table. The correct result is obtained within
a similar duration.
```
Limit (cost=2700.68..2700.70 rows=8 width=29) (actual time=12.561..12.568 rows=8 loops=1)
-> Sort (cost=2700.68..2701.50 rows=328 width=29) (actual time=12.559..12.560 rows=8 loops=1)
Sort Key: (count(*)) DESC, (((res_partner.display_name)::text ~~* '%eezee-i%'::text)) DESC, res_partner.display_name
Sort Method: top-N heapsort Memory: 26kB
-> GroupAggregate (cost=2687.56..2694.12 rows=328 width=29) (actual time=12.449..12.527 rows=56 loops=1)
Group Key: res_partner.id
-> Sort (cost=2687.56..2688.38 rows=328 width=20) (actual time=12.408..12.425 rows=305 loops=1)
Sort Key: res_partner.id
Sort Method: quicksort Memory: 42kB
-> Nested Loop Left Join (cost=1283.21..2673.86 rows=328 width=20) (actual time=11.629..12.340 rows=305 loops=1)
-> Bitmap Heap Scan on res_partner (cost=1282.79..1618.17 rows=328 width=20) (actual time=11.596..11.862 rows=56 loops=1)
Recheck Cond: (((email)::text ~~* '%eezee-i%'::text) OR ((display_name)::text ~~* '%eezee-i%'::text) OR ((ref)::text ~~* '%eezee-i%'::text) OR ((vat)::text ~~* '%eezeei%'::text))
Rows Removed by Index Recheck: 1
Filter: (active AND (((type)::text <> 'private'::text) OR (type IS NULL) OR (type IS NULL)))
Rows Removed by Filter: 1
Heap Blocks: exact=58
-> BitmapOr (cost=1282.79..1282.79 rows=328 width=0) (actual time=11.560..11.560 rows=0 loops=1)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..322.22 rows=162 width=0) (actual time=4.724..4.724 rows=54 loops=1)
Index Cond: ((email)::text ~~* '%eezee-i%'::text)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..322.22 rows=162 width=0) (actual time=3.825..3.825 rows=56 loops=1)
Index Cond: ((display_name)::text ~~* '%eezee-i%'::text)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..321.00 rows=1 width=0) (actual time=1.752..1.752 rows=0 loops=1)
Index Cond: ((ref)::text ~~* '%eezee-i%'::text)
-> Bitmap Index Scan on res_partner_name_tgm_idx_gin (cost=0.00..317.03 rows=4 width=0) (actual time=1.254..1.254 rows=0 loops=1)
Index Cond: ((vat)::text ~~* '%eezeei%'::text)
-> Index Only Scan using sale_order_partner_id_index on sale_order (cost=0.42..2.74 rows=48 width=4) (actual time=0.006..0.007 rows=5 loops=56)
Index Cond: (partner_id = res_partner.id)
Heap Fetches: 12
Planning time: 1.056 ms
Execution time: 13.791 ms
```
TaskID: 2031147
Co-authored-by: Yannick Tivisse <yti@odoo.com>
Instead, make sure the field src is always up to date
Change a bit the behaviour of `_write` to always try to update potential
translations when in monolanguage (was only done in multi-language if different
than en_US) and always create translations when writing on a translatable field
when in multi-language (was skipped in en_US)
Add tests
Task-id: 2031752
Pad: https://pad.odoo.com/p/r.dbf835d0aae979d858f73446dcbfb1dfclosesodoo/odoo#34598
Signed-off-by: Martin Trigaux (mat) <mat@odoo.com>
If a db is in single lang (en_US or not) but still has translations, the
translation should be updated in addition to updating the referenced record
Add tests to formalise the expected behaviour:
If one language only (en_US or not), when writing on a translatable field:
- record field should updated
- value of potential existing en_US translation should be updated
- src of potential existing translations should be updated
- no new translation should be created
If en_US and fr_FR, when writing on a translatable field in en_US:
- record field should updated
- value of potential existing en_US translation should be updated
- src of potential existing translations should be updated
- new en_US translation should be created if was not present
If en_US and fr_FR, when writing on a translatable field in fr_FR:
- record field should not be updated
- value of potential existing fr_FR translation should be updated
- src of potential existing translations should not be updated
- new fr_FR translation should be created if was not present
Adapt test_new_api test
get_installed is ormcached, just putting active = True is not enough
Instead, make sure the field src is always up to date
Add tests
Changes in _write:
- Replace _set_ids (to be deprecated) call by _upsert_translations as it works
in batch
- Call _upsert_translations for any language, including en_US
In case an English translation already existed for a record, only the master
value (on the reference record) was updated but the user kept seeing the
translation value (was revealed 73a7534bfc).
- Read src_trans without language
Similar as above, if an English translation already existed, the translation
was used for the result of the read and not the new value that has just been
inserted into the database
- Add _set_source method
When updating a master record of a translated field, the src field must be
updated, including in different languages.
Before it was ok that the src field was out of date as the source was computed
Changes in copy_translation:
- set src as the new value without lang
update the comment to reflect reality since 489494e733
src will contain the English version if no changes were made and it will
contain the modified value if copy was overriden
Changes in upsert_translations:
- Do not force a module, comment, state
Only src, res_id, name, value and lang must be given. Optional values will no
longer be set to null if not given
This class referenced demo data but even if they weren't installed, the
tests were passing. Remove the definition and usages.
closesodoo/odoo#35343
Signed-off-by: Simon Lejeune (sle) <sle@openerp.com>
Reproduce:
1. I create a product with fifo and real_time valuation (make sure the invoicing policy is 'delivered' and make it available in pos).
2. set the cost of the product to be 5.0 and sale price to be 10.0.
3. I update its inventory and set 5 items. -> total valuation of 25.
4. I change the cost of the product to 1.0.
5. I update again the quantity to 10, making the valuation of the product to 30.0.
6. I configure point of sale to allow invoicing.
7. Open a pos session then sell 7 items of the product (invoice the order).
8. Since the product is fifo, this means that the cost of goods sold is 5 * 5.0 + 2 * 1.0 = 27.0.
9. The correct expense account line should have balance of 27.0.
10. The correct output account line should have balance of -27.0
However, instead of 27.0 as stock valuation amount, the value in the
pos.order invoice is currently 7 * 2.0 = 14 -- and this is wrong. It is
not taking into account the value of the previous stock
Solution:
Missing refactoring from https://github.com/odoo/odoo/commit/14812f41710282f9923078f6764323086500ae85closesodoo/odoo#35335
Signed-off-by: Simon Lejeune (sle) <sle@openerp.com>
Since #34996, in some case, the error detail was not logged keeping
only the final summary "x errors, x failures" when running TestSuite.
Making fail test test_cache_invalidation for instance won't show
a detailed message when breaking test_01_project_tour will.
This issue occurs when using subtest, like with assertQueryCount,
@users decorator, test_all_l10n, and test_youtube_urls.
Since Testresult addSubTest append directly to error and failures
instead of calling addError and addFailure, we need to ovewrite
addSubtest too.
closesodoo/odoo#35270
Signed-off-by: Denis Ledoux <beledouxdenis@users.noreply.github.com>
The "company_id" fields for the iap account has
been changed by the "company_ids" fields to allow
the use of an iap account with multiple company.
closesodoo/odoo#35093
Signed-off-by: Rémi Rahir (rar) <rar@odoo.com>
qmo-odoo