Coverage and test runs often result in JavaScript heap out of memory

[arr00]

On MacOS, vanilla runs of the test and coverage npm scripts often results in a JavaScript heap out of memory error. This is a known issue with hardhat NomicFoundation/hardhat#3471.

Increasing the allowed old memory to using the following env var fixes the issue NODE_OPTIONS="--max-old-space-size=8192". We should consider documenting this or even setting the env var by default.

[Amxx]

The .github/workflows/checks.yml contains

env:
  NODE_OPTIONS: --max_old_space_size=8192 

On my machine, I have it set in ~/.env, which is then sourced by ~/.zshrc.

I honestly think this is a very frequent error that is very well documented online. Should we make that default for everyone ? If we do so, could that cause issues on machine that have smaller ram ?

[Amxx]

For the record, if I type "JavaScript heap out of memory" in google, the first link I get is this, which contains all the information to solve that.

[arr00]

I think we should set it by default--I'm not sure if it would cause issues on machines with too little memory... but the run would fail regardless for them.

[Amxx]

How would you set it ? Shipping a .env is considered bad practice

[arr00]

We could prepend the npm scripts in package.json like so:

"test": "NODE_OPTIONS=\"--max-old-space-size=8192\" hardhat test"