new integration test: Blocking users with low followers

[dunn]

The way I did this requires four new env variables:

• BT_TEST_LOW_FOLLOWER_USER (a user with < 15 followers)
• BT_TEST_LOW_FOLLOWER_PASS
• ACCESS_TOKEN
• ACCESS_SECRET

The last two are so we can use the Twitter API directly to unblock the low-followers user after the tests, since doing it through the web UI via Casper didn't work.

[jsha]

Instead of ACCESS_TOKEN and ACCESS_SECRET in the env, how about pulling them from the DB after logging in, as in unblock-all.js? As-is, it's not clear which user the ACCESS_TOKEN belongs to, and it may get unwieldy to manage with more tests.

[dunn]

Oh, even better, we can just run unblock-all.js.

[dunn]

Oops, these passes are false positives: https://travis-ci.org/dunn/blocktogether/jobs/159382890#L1315

[jsha]

Overall, looks great - thanks for continuing to work on these tests!

[dunn]

Most of your comments are about doing things via the API rather than through Casper.

I forgot that at least some Node modules can be required in Phantom (http://docs.casperjs.orgf/en/latest/writing_modules.html), so I'll take a look at doing that.

[jsha]

If we can't pull in the modules directly we could probably shell out to executables, right? I.e. We can tweet by shelling out to jwurl with the right url (and possibly adding post support?)

On Sep 13, 2016 18:21, "Alex Dunn" notifications@github.com wrote:

Most of your comments are about doing things via the API rather than through Casper.

I forgot that at least some Node modules can be required in Phantom ( http://docs.casperjs.orgf/en/latest/writing_modules.html), so I'll take a look at doing that.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/jsha/blocktogether/pull/231#issuecomment-246877713, or mute the thread https://github.com/notifications/unsubscribe-auth/AANcLX_N8uZTQNf4nPwfZ1BqZLKRH9hhks5qp0wWgaJpZM4J6IVj .

[dunn]

I don't think so, since Casper can't do arbitrary code execution and the timing is important. But I'm pretty sure we'll find compatible modules in order to access the API.

[dunn]

For some reason the tests are acting up locally, so I might have to test with Travis for a bit; if you see a lot of small commits being pushed to this branch that's why.

[dunn]

Sorry for the delay; I'm looking at using the API functions and the Twitter clients are pulling in Node-specific stuff.

[dunn]

Now it's working locally and failing on Travis... (https://travis-ci.org/dunn/blocktogether/builds/160610253)

[jsha]

I'm going to go ahead and merge this - better to have the test than not, and we can work on improving it later. Thanks again!

[jsha]

Oh, I see you did move to using unblock-all to unblock through the API after all! Sorry, I had missed that on my last read through.

[dunn]

Yeah, I just couldn't figure out how to do the operations within the test via the API.