Looks like one of the libraries (not sure which one) updated its bower.json file to use the master version for one of its dependencies, which then brought in a non-es-modules transitive dependency.

The last commit printed the contents of Data.HeytingAlgebra.js to see whether it's using CommonJS modules or not (just as a sanity check). It's using ES6, so at least caching isn't the issue.

Looks like 82467e7 is building fine again. In that commit, I updated prelude to point to the state of es-modules branch before I added the commits in purescript/purescript-prelude#284. That branch is the fix-es-modules branch.

The latest commit updates prelude to point to fix-es-modules--change-unit, which is the same as fix-es-modules but with another commit that changes unit's FFI from {} to undefined. If that doesn't cause CI to fail, then I'm wondering if these do:

- exports.var functionName = function() { ... };
+ export const functionName = function() { ... };

Perhaps we need to use this version to fix things?

export function functionName() { ... };

The Ubuntu build builds fine even though unit is changed from {} to undefined. So that leaves the export const functionName question.

Looks like using export const functionName = function() { .. } is causing the CI errors. @sigma-andex Thoughts on this?

There's nothing wrong with that unless it's just a style lint thing, which can be disabled.

There's nothing wrong with that unless it's just a style lint thing, which can be disabled.

Could you clarify this? I'm not sure what you mean. If you mean disabling an eslint thing, I don't think that's the issue, but I'm guessing that's not what you mean. AFAICT, the CI errors are caused by purs thinking that the exports of Data.HeytingAlgebra's FFI file, which uses export const rather than export var, are invalid.

the exports of Data.HeytingAlgebra's FFI file, which uses export const rather than export var, are invalid.

Ignore my comment then. The issue is that this branch changed exports to const, but the interim ESM bundler which hasn't been removed yet doesn't understand it. My suggestion would to revert any such changes on those branches, and remove the bundler before doing any other work on those branches.

Edit: I think you came to that conclusion elsewhere 😆

I think you came to that conclusion elsewhere 😆

😁

Succeeded by #4255