The common editor layout¶
Create the recipe¶
You can create a recipe:
From the Flow tab, by clicking in the toolbox
From the Recipes tab, by clicking on the New recipe button
From the actions menu of the dataset
From the “More” menu of a dataset
If you have not already selected one, the first step when writing a code recipe is to select the datasets that are used as “inputs” of your recipe. You may only read in these datasets, not write.
Click on the “+” button for each dataset that you need.
In the above screenshot, a first dataset “www_dataset” has been selected as input.
You then need to select or create the output datasets. Generally, when you create a recipe, you will be creating its output dataset at the same time. Most times, the output datasets of a recipe will be managed datasets (for more information on Managed datasets, see the DSS concepts page).
- Give a name to the output dataset
- Select in which connection it will be stored. For more information about the concept of storing Managed Datasets into connection, see DSS concepts
- You may choose to copy the schema of one of the input datasets or to use a “brand new” schema for your output dataset.
You will almost always need a schema for the output dataset, but many recipes can help you and auto-fill the schema of the output datasets.
For example, when you write a “SQL” recipe (where you write a SQL query to create a dataset), the columns returned by the SQL query can automatically become the columns of the output dataset schema.
When you click the “Create and use” button, a new empty dataset will be created in the hdfs-managed connection.
Once you have selected input and outputs, you can tune the name of recipe and save.
You can start writing your code in the “Code” tab.
The code should fill data in the output datasets. Please refer to the specific documentation for each recipe for more information about how to do that.
Run the recipe¶
Code recipes have a “Run” button that automatically appears as soon as you have defined at least one output dataset for the recipe.
When you click the Run button, a new job is started. When it’s over, you get either a success or error message and can explore the generated output datasets.