E2E testing via Continuous Integration?

[lowlyocean]

Hi folks, I'm not really sure if there's a channel in Discord where devs can discuss Plaid in real-time, but I think this is an appropriate place for this question:

If someone's built an integration with Plaid, such as this Pattern project, what facility is there for end-to-end testing as part of the CI/CD pipeline? For example, I want to set up canned account/transaction data as part of my tests and ensure some button press in the UI renders correctly my app's transformations on that account/transaction data.

Is there a self-contained stateful mock of the Plaid API (something like this for Stripe)? Or would my pipeline need to really connect to Plaid's servers to use some Sandbox? Any active examples of E2E testing a Plaid integration in CI/CD pipeline would be appreciated! Thanks

[phoenixy1]

@lowlyocean there is! fill out this form https://forms.gle/HQeHvZbWvsL1eBrEA and I'll invite you to the Discord and let's chat about this question there!

[hxlnt]

@phoenixy1 @lowlyocean Was there an update or outcome on this that could be shared outside of Discord? I appreciate that GitHub might not be the right place for discussion, but I do think that the mock data-for-pipeline question is great, and the Plaid team's recommendations could benefit many of us working with the API. Thanks!

[phoenixy1]

@hxlnt For account and transaction data in particular, there are two ways to test it:

First, you can use Plaid's custom users (https://github.com/plaid/sandbox-custom-users/). These are good for ensuring totally deterministic data, which is useful for test suites, but you also have to generate the test data yourself (we do provide some pre-populated templates for you) and you don't have full control over when transactions updates occur, since these users emulate the real-world behavior in which Plaid controls the schedule of transactions updates.

Alternatively, you can use the user_transactions_dynamic test user (https://plaid.com/docs/transactions/transactions-data/#testing-pending-and-posted-transactions) on Sandbox. This user generates data that is somewhat randomized and not fully deterministic; the advantage is that it more closely mimics real-world behavior and that Plaid generates the data for you. This test user also gives you tighter control over when and how new transactions and updates are generated, because you manually trigger the process (by calling /transactions/refresh), as opposed to the custom user data, which generates transactions updates based on the dates input into the file.

I'm not aware of any open source projects using Plaid that also have transactions test suites built out.

Hopefully that answered your question, but let me know if it doesn't!