You should understand how your account's API version determines the requests allowed by an API and the responses generated. The version also determines the structure of events generated by API requests.
You should always ensure that your API keys are configured for the version expected by your code. In other words, when your code is deployed from test to production, the version on the keys should match the code version.
Digital River often makes non-breaking changes to our API request and response content. As a result, we recommend your integration conform to the tolerant reader principle. Specifically, this means that you should:
Be aware that new elements can be added to responses at any time.
Build your code to extract only the attributes needed when reading responses and ignore everything else.
Avoid coding with a specific order of fields in mind.
Assume that ids are alphanumeric strings, which potentially contain special characters, and they have variable lengths.
Expect changes to the length and value of error messages and other strings that don’t represent an enumeration, type, or code.
Anticipate the addition of new optional request and query parameters.
You should be aware of how dates and times in the Digital River API are represented and ensure they are properly formatted in your requests.
To improve fraud detection, you should provide an IP address when creating a checkout, invoice, or order.
To make call debugging easier, you should use specific HTTP request headers.
Attempt to minimize HTTP
400 Bad Request and
409 Conflict error types by adding appropriate validation checks before a request is submitted.
GET requests, we encourage making concurrent calls.
Avoid making changes to the same resource in multiple calls. Instead, bundle changes in a single call.
Avoid making concurrent mutation calls to the same resource.
You should use the
liveMode flag contained in API responses to validate that you are pointing to the correct environment.
The webhook end point must be able to handle concurrent webhook callback requests.
Webhook events may be delivered multiple times. So be sure you can process the delivery of duplicate events.
Your webhook endpoint must respond to callback requests in a timely manner. A response time greater than 3000 milliseconds is considered a timeout. We expect you to immediately acknowledge the callback request by sending an appropriate HTTP
2XX response code. Once this acknowledgment is sent, you can then asynchronously process the webhook event on your end.