ai16z / eliza

Conversational Agent for Twitter and Discord
https://ai16z.github.io/eliza/
MIT License
2.43k stars 704 forks source link

Tests are failing #64

Open sirkitree opened 1 month ago

sirkitree commented 1 month ago

Tried out running the tests today (npm run test) and there are a lot of failures currently.

We should

  1. evaluate which tests are still necessary
  2. get them working
  3. setup automated github action to run test for every pull request
Node.js v22.8.0
Test Suites: 16 failed, 16 total
Tests:       58 failed, 58 total
Snapshots:   0 total
Time:        14.737 s
Ran all test suites.

@lalalune could you advise here? I think we can work on the tests but the first task of identifying which ones are necessary would be great if you could help with that more quickly than others.

sirkitree commented 4 weeks ago

from @lalalune

there are some, but they need a little love i had tests in the old version: https://github.com/jointhealliance/bgent https://github.com/JoinTheAlliance/bgent/tree/main/src/lib/__tests__ https://github.com/JoinTheAlliance/bgent/tree/main/src/lib/actions/__tests__ https://github.com/JoinTheAlliance/bgent/tree/main/src/lib/evaluators/__tests__ https://github.com/JoinTheAlliance/bgent/tree/main/src/lib/providers/__tests__ these are the old ones i moved most of these into tests if you wanna get the test automation working with a hello world test, i can assign this

sirkitree commented 4 weeks ago

After the above is merged, we need to quantify existing tests and decide which are still relevant.

I'll analyze the test files and describe their purposes. Here are the test suites found in the codebase:

Basic Tests

Memory Tests

Browser Service Tests

Token Provider Tests

Goals Tests

Runtime Tests

Actions Tests

Continue Action Tests

Messages Tests

Evaluation Tests

The test suite uses Jest as the testing framework and includes features like:

Most tests include proper setup and teardown procedures to ensure a clean testing environment for each test case.

sirkitree commented 3 weeks ago

Now that we have a basic hello world type of test in place and running on ci, we can start lookin at these other ones. I'm going to split each one of these into their own task so they can be handled piece by piece and possibly spread out amongst the collective.