This is probably not enough of a margin (you need time for the child process to get killed), but I get the idea

We still want an error if a timeout occurs, so exit(1) or something like that.
Otherwise if the timeout occurs the CI will return no error.

you can do a single message with both infos, not sure what the benefit is of having 2 messages

only test this one during draft review

we use before, which runs before all tests, to get data that will be used in all tests (rpc_timeout and chronik_url). We set the timeout for all tests in this file and the chronikUrl var that will be used by all tests in this file

we need to use async function() {} instead of async () => {} to correctly access mocha's this and set the timeout

this is a to-do for all mocha tests...apparently arrow functions are bad practice because we can't access this. First time I have needed to access it.

https://stackoverflow.com/questions/23492043/change-default-timeout-for-mocha

I could not find a specific way to trigger a callback in the event of a mocha timeout. However, looks like they work the same way as any other mocha failure. e.g. if I set the timeout to 1s and run this test, here's the output:

> chronik-client@0.23.0 integration-tests
> mocha -j1 -r ts-node/register test/integration/block_and_blocks.ts

Starting test_runner for chronik-client_block_and_blocks
Test runner for chronik-client_block_and_blocks started


  /block and /blocks
2024-02-26T23:52:25.607000Z TestFramework (INFO): Initializing test directory /home/joey/github/abc/bitcoin-abc/build/test/tmp/test_runner_₿₵_🏃_20240226_155225/setup_scripts/chronik-client_block_and_blocks_0
    1) "before all" hook for "gives us the block and blocks"
Tests complete, shutting down child process


  0 passing (1s)
  1 failing

  1) /block and /blocks
       "before all" hook for "gives us the block and blocks":
     Error: Timeout of 1210ms exceeded. For async tests and hooks, ensure "done()" is called; if returning a Promise, ensure it resolves. (/home/joey/github/abc/bitcoin-abc/modules/chronik-client/test/integration/block_and_blocks.ts)
      at listOnTimeout (node:internal/timers:573:17)
      at processTimers (node:internal/timers:514:7)



Setting chronik url to  http://127.0.0.1:30001
rpc_timeout for block_and_blocks: 60
Mocha timeout set to 240s
2024-02-26T23:52:26.722000Z TestFramework (INFO): Passed test setup data to mocha
2024-02-26T23:52:26.722000Z TestFramework (INFO): Received a stop message, exiting
2024-02-26T23:52:26.773000Z TestFramework (INFO): Stopping nodes
2024-02-26T23:52:27.026000Z TestFramework (INFO): Cleaning up /home/joey/github/abc/bitcoin-abc/build/test/tmp/test_runner_₿₵_🏃_20240226_155225/setup_scripts/chronik-client_block_and_blocks_0 on exit
2024-02-26T23:52:27.027000Z TestFramework (INFO): Tests successful
Running Unit Tests for Test Framework Modules
setup_scripts/chronik-client_block_and_blocks.py started
setup_scripts/chronik-client_block_and_blocks.py passed, Duration: 2 s

TEST                                             | STATUS    | DURATION

setup_scripts/chronik-client_block_and_blocks.py | ✓ Passed  | 2 s

ALL                                              | ✓ Passed  | 2 s (accumulated) 
Runtime: 2 s

mb another failure that we get setup_scripts/chronik-client_block_and_blocks.py passed, Duration: 2 s ... but this does confirm the process is killed right?

we move this setup, which applies to the whole test, out of before

In this test specifically, the step numbers are confusingly named. We have a test for the above yield True which tests initial regtest before 10 blocks are mined. Not directly related to this improvement, but since we kind of add a "step 0" with setup, was difficult to not correct this here.

we add a zero step to send the timeout (and chronik info), which is used in every test

If you want your test to be relevant, add:

import time

time.sleep(10) # wait 10 seconds

in the setup script. Make sure to sleep more than the mocha timeout so the script still runs when it occurs.

OK, this should be automated by the setup framework at some point because it will be identical for all the setups

The step 0 should be moved out from the scripts and moved to the class so it's automated. Something like the setup_nodes or setup_network from the test framework that can be overridden by the child script class if needed. Maybe setup_test_info ?

all tests have at least 1 node, it's fine to use self.nodes[0] and let the script override this if needed