Routing tasks for shadow running

# Routing tasks for shadow running We will need to handle three different events. ## Order Events ### Event: Created Order -> Import Order Command Handler - Retieve Warehouse(CP -> WarehouseID). As first approach we can have hardcoded the CPs for the evalutaed warehouses. - Find Dataset in Repo (WarehouseID, Date) - Only non **final** datasets can be used. Since final datasets are static ones that could have been used by static executions, making the results inconsistent. - [Dataset NOT exists] - Create Dataset - The dataset will only contain the order received. - Create Config. - We may have a configuration already for this ?. We probably want to create a fresh new one. In any case we must be using `tags` since we must be sure that this configuration is not shared among different days/warehouse, since this will provoke being modifying an unexpected execution. - This must be a defaut configurarion with only one vehicle. - Create Execution - With (Dataset, Config, Algorithm). We can use by default Plotwise. - [Dataset Exists] - Update Dataset - Build new order based on the info provided by the cmd. - Add Order to the dataset. - *[Maybe]* Update config based on the number of vehicles per number of orders - We need to identify a non final configuraiton by (dataset, warehouseID), we need again tagging as a way to locate the configurations. Then, we need to be sure that the config is in dynamic state. - Update execution. - Send the new Order to Plotwise. ### Event: Updated Order -> Update Order Command Handler - Find Dataset using OrderID. **Possible performance issues since we are storing all the orders as a list of JSONs**. - Check if the updated order still belongs to the same dataset - [Belongs to the same dataset] - Update dataset. - Send order update to the algorithm. - [Plotwise] Updating an order for plotwise require to delete the old one and a new one. - [Belongs to other dataset] - Remove order from old dataset - Update algorithm executions removing the deleted order. - Add order to new dataset/execution: - Find Dataset. We will repeate the process for a created Order. So maybe it may makes sense to call the **ImportOrderCommandHandler**. The main problem with this approach is that we need to share the transaction among the commamd Handlers. ### Event: Deleted Order - Find Dataset using OrderID - Update Dataset -> Remove order from list of orders - Update Algorithm execution -> Remove order from the continuos execution ## Parcel Events ### Parcel Created - Find dataset by OrderID - Update Dataset. - Update Algorithm Execution. Again we are going to need to remove the old order/event and a add a new one with the package. ### Parcel Updated. - Find dataset by OrderID - Update Dataset. - Update Algorithm Execution. Again we are going to need to remove the old order/event and a add a new one with the package. ### Parcel Deleted. - Find dataset by OrderID - Update Dataset. - Update Algorithm Execution. Again we are going to need to remove the old order/event and a add a new one with the package. ## Database changes - Do migrations on dataset to allow `dynamic` flag - Do migrations on config to allow `dynamic` flag - We need to add tags to configs in order to not share configurations among different continuos executions for differents days. - Supplementary table for plotwise to know the event associated for an order, in case of having to update/modify the order. # Using default configuration: Routing tasks for shadow running ## Simplications - Using a hardcoded configuration We will need to handle three different events. ## Order Events ### Event: Created Order -> Import Order Command Handler - Retieve Warehouse(CP -> WarehouseID). As first approach we can have **hardcoded** the CPs for the evalutaed warehouses. - Find Dataset in Repo (WarehouseID, Date) - Only non **final** datasets can be used. Since final datasets are static ones that could have been used by static executions, making the results inconsistent. - [Dataset NOT exists] - Create Dataset - The dataset will only contain the order received. - Create Config. - Create Execution - With (Dataset, Hardoded Config, Plotwise Algorithm). - [Dataset Exists] - Update Dataset - Build new order based on the info provided by the cmd. - Add Order to the dataset. - Update execution. - Send the new Order to Plotwise. ### Event: Updated Order -> Update Order Command Handler - Find Dataset using OrderID. **Possible performance issues since we are storing all the orders as a list of JSONs**. - Check if the updated order still belongs to the same dataset - [Belongs to the same dataset] - Update dataset. - Send order update to the algorithm. - [Plotwise] Updating an order for plotwise require to delete the old one and a new one. - [Belongs to other dataset] - Remove order from old dataset - Update algorithm executions removing the deleted order. - Add order to new dataset/execution: - Find Dataset. We will repeate the process for a created Order. ### Event: Deleted Order - Find Dataset using OrderID - Update Dataset -> Remove order from list of orders - Update Algorithm execution -> Remove order from the continuos execution ## Parcel Events ### Parcel Created - Find dataset by OrderID - Update Dataset. - Update Algorithm Execution. Again we are going to need to remove the old order/event and a add a new one with the package. ### Parcel Updated. - Find dataset by OrderID - Update Dataset. - Update Algorithm Execution. Again we are going to need to remove the old order/event and a add a new one with the package. ### Parcel Deleted. - Find dataset by OrderID - Update Dataset. - Update Algorithm Execution. Again we are going to need to remove the old order/event and a add a new one with the package. ## Database changes - Do migrations on dataset to allow `dynamic` flag - Do migrations on config to allow `dynamic` flag - We need to add tags to configs in order to not share configurations among different continuos executions for differents days. - Supplementary table for plotwise to know the event associated for an order, in case of having to update/modify the order. # Using executions as entrypoint A new order arrived. 1. Look for WarehouseID 2. Look for executions(WhID, date) 3. [Found Executions] 4. Create new dataset. 5. Create new config -> Hash() (date, warehouse) 6. Create new execution. 5. [Not Found Executions]