https://github.com/openfoodfoundation/openfoodnetwork/issues/2007
Spree 2.0.4 creates in-memory shipments for the payment step. That means
we don't need to create a shipment in that transition. This change
copies the checkout_flow from the Spree code, but leaves one last
customisation: removal of the confirm step.
75ec77dc31
We are adding taxons to the product as you change the primary taxon.
However we never remove the previous primary taxon so it forces the user
to update the taxons manually. This can be a big problem if you are bulk
updating products.
We now remove the taxon that matches the previously set primary taxon.
Working on the Spree upgrade, we found that this scope is using the soon
obsolete column `spree_products.count_on_hand`. Trying to measure the
impact of changing this scope, I couldn't find any use of it.
There is a variable called `active_distributors` used when serialising
enterprises, but that variable is initialised with
`Enterprise.distributors_with_active_order_cycles.ready_for_checkout`,
not using the `active_distributors` scope.
See also:
https://github.com/openfoodfoundation/openfoodnetwork/issues/2013
* master: (206 commits)
embedded groups layout changes
embedded groups initial test
Allow html requests for OrderCycleController#update
Move applicator calls to OrderCycleForm
Refactor OrderCycleForm to make logic clearer
Extract schedule syncing logic into OrderCycleForm
Add basic specs for OrderCyclesController#create
Add basic OrderCycleForm to handle create/update logic
Remove unnecessary respond_to blocks from OrderCyclesController
Include admin users as managers on new enterprises
Remove obsolete goWithoutHashFragments
Simplify Navigation.go, not preserving hash fragments
Only show change warning for open order cycles
Use a SubscriptionsCount query object to provide counts to IndexOrderCycleSerializer
Preload subscription counts for serialization in order cycle collection actions
Request the subscription count for change warning each time, don't cache
Ask user to confirm oc date change for open order cycles with subsciptions
Fix ordering of Gemfile.lock
Add rack-rewrite to handle redirects
Renames product bulk edit action to index
...
It was introduced, because `Navigation.go` perserved hash fragments. We
actually don't need that behaviour and it has been corrected.
`goWithoutHashFragments` also didn't deal with absolute URLs. And it used
`location.origin` which is not supported by Internet Explorer. That is
fixed by our use of Modernizr though.
I looked through the history and it looks like this function was a bit
flawed (preserving hash fragments) from the beginning. It has been
patched a few times without addressing the underlying issue that we want
more than just replacing the pathname. We want to go somewhere else.
