Note that, as explained in
https://apidock.com/rails/v3.2.13/ActiveRecord/Relation/delete, `delete` does
not trigger callbacks and so it skips the products cache logic.
If we still want to avoid instantiating the AR object, we need to explicitly
call that logic for the cache to be up-to-date.
A previous pull request added support for flexible decimal characters
when editing money amounts.
https://github.com/openfoodfoundation/openfoodnetwork/pull/1831
This pull request applies the same principle to the weight calculator
which was missed in the previous pull request.
The admin variants controller create action is adapted (same as in the admin ProductsController). This is necessary because variant.on_hand and on_demand cannot be mass assigned on creation in v2, see VariantStock for more details.
For now we are hiding the new stock management page that comes in spree v2 as we don't need the complexity added by the management of stock per stock location (only one stock location in ofn v2 for now)
We keep tests running by default without cache, keeping it as it was before 10a79d5a65
Additionally, cached_products_renderer no longer depends on Rails.env, so here we remove that dependency from it's spec.
This way we don't need to touch the class implementation to enable the
products cache in development. Just change the default value in
`app/models/spree/app_configuration_decorator.rb`.
This separates logic for bang and non-bang versions of
Spree::Order#next.
The different conditions used in both methods (state == "completed" vs
order.completed?) have implications in whether a transition is attempted
or not.
The bulk order management page tried to wait for orders being loaded.
But instead of receiving a request object with an a promise to wait for
the page received a simple array without promise. As a result, the page
didn't wait for orders being loaded.
When loading of orders took longer than loading of line items they or at
least not all of them were enriched with distributors and order cycles
and the line items were missing some orders. That lead to random spec
failures.
A spec has been added to check that the attributes for the order states
after "cart" ("address", "delivery", "payment") are retained if the
order is transitioned to completion, BUT currently this is passing
because it is the only shipping method available.
