Nock VS pact-js

So, why would you want a REAL server to mock request/reponses? You have a lot of intercepts today that sit on the network layer and you can define things like "If you send request to that endpoint, with that json, please return that Status" (for NodeJS example, Nock - https://github.com/nock/nock)

That said, folks seem to like "recording" features in these sorts of tools (Ruby's VCR, nock, etc), so maybe there's a future where I add something similar. I've always just found the ergonomics of those features awkward to deal with, especially having to flip back and forth between tests and fixtures files to figure out what's wired to what, but maybe there's a clean solution... perhaps a "live request" mode that just prints mock code snippets of request/response pairs passing through your app.

While not mocking per say I usually use nock for http calls. You can use nock.recorder.rec() to capture the http call to play back during test, That way you are always using "live" code but not making real calls to servers.

I looked at the service which is used to get the feed URLs from a blog URL and noticed it takes the html response of the blog URL and gets the links ( tags) by checking the type attribute value against a list of valid feed values. So, I decided to use a similar approach by getting the html response for a provided URL and checking the Content-Type header against a list of valid MIME types for a feed. I ended up updating the logic to test if a URL is a feed URL, returning it if true. If the URL is found to not be a feed URL, it would try to get the feed URLs assuming the URL is a blog URL. I tested and confirmed that the new logic worked for both blog and feed URLs. Then, I added some tests for the new function I added to test for a feed URL. Testing this ended up being simpler than I expected as all I had to do was mock the response of a test url (using nock), and then check if the function returned the correct boolean value for a url. I created a PR and noticed that some of the tests in another file were now failing. While I was investigating this, I got a review on my PR, requesting me to add another test to the file which had the failing tests. That file tested the API service as a whole. I found out that nock only mocks a URL's response for one request by default. And since I was now checking for a feed URL as well, the function which returned the feed URLs from a blog URL was throwing an error since the nock for that was used up. To fix this, I had to specify in the nock statement to mock the URL response for two requests:

There are libraries like https://github.com/nock/nock to prevent mocking the whole axios.

Use nock: https://github.com/nock/nock

Going based off the documentation I sent you in my last reply, there is an example that uses nock to emulate api responses. I haven't used nock myself, but the example seems pretty simple to use. You just need to take the example and change the response object to be the shape of what your getStuffFromDatabase function returns. That way your useCategory function runs as close to normally as possible, while providing a mock response value instead of hitting the database.

If you’re willing to venture into nodejs territory, then nock is a fantastic and simple to set up http mock server. https://github.com/nock/nock

First, he shows the official pact.io websites. Then, he clicks on the “View on Github” button by selecting Node JS and Javascript from the list of options.

It just just enough abstraction over the basics of converting your HTTP calls to GraphQL queries and mutations, but ALL of the parsing of responses is on you. I’m well versed in parsing JSON in Elm. I’m also familiar with the compiler errors as well as runtime errors you get with JSON that doesn’t match up to what you designed. At some point I’ll probably have to move beyond the unit tests and add contact tests, maybe via Pact.js.

> I also wasn't aware that "unit" referred to an isolated test, not to the SUT.

I'm with you. That claim is unsubstantiated. It seems to trace to the belief that the first unit tests were XUnit family, thus were SUnit for Scheme. But Kent Beck made it pretty clear that SUnit "units" were classes.

https://web.archive.org/web/20150315073817/http://www.xprogr...

There were unit tests before that. SUnit took its name from common parlance, not vice versa. It was a strange naming convention, given that the unit testing framework could be used to test anything and not just units. Much like the slightly older Test Anything Protocol (TAP) could.

> [on unit tests] This does lead to a lot of work maintaining them whenever the implementation changes, but this is a necessary chore because of the value they provide.

I disagree. Unit tests can still be behavioral. Then they change whenever the behavior changes. They should still work with a mere implementation change.

> This is why I still think that the traditional test pyramid is the best model to follow.

I'll disagree a little with that, too. I think a newer test pyramid that uses contract testing to verify integrations is better. The notion of contract tests is much newer than the pyramids and, properly applied, can speed up feedback by orders of magnitude while also cutting debugging time and maintenance by orders of magnitude.

On that front, I love what Pact is doing and would like to see more competition in the area. Hottest thing in testing since Cypress/Playwright . . .

https://pact.io

I've worked in places where Pact [0] was used for testing services developed by different teams (external) and teams themselves (internal)

[0] https://pact.io/

I'd never heard anyone call those "integrity tests" before. I think "contract test" is more common.

Assuming I understood you, that is.

I've been telling everyone to look at Pact to make contract testing easier to organize and maintain and to make it easier to trigger in the other tests in CI when an interface's behavior changes. They haven't offered me a commission yet. ;-)

https://pact.io

We use pact and since introducing it we have significantly increased velocity and reduced test cycles as it catches things very early. For system tests we hand write them using whatever test frameworks the team is used to.

There is also Pact https://pact.io/ for a language agnostic pact testing.

When you wish to focus on the contract ( which kind of field is required, ...), you shoud use contract testing frameworks. As you seem to leverage a microservices, a consumer driven contract testing approach with a framework like Pact.js is recommended.