Mars Hall

Also, the `resource "heroku_build" "middleware-frontend-build"` should not have that `depends_on`, because the `app_id` already contains a dynamic reference to that parent resource.: ```hcl resource "heroku_build" "middleware-frontend-build" { app_id = heroku_app.middleware-frontend-app.id...

Also worth noting, your linked gist of the Terraform log does not contain any app creation, nor setting of config vars. So there's nothing in that log to indicate that...

Any update @ArtOfFugue66 ?

> make stderr the default While this might be a breaking change for folks that have been forced to parse stdout for results of a command, it would ensure that...

`fetch` doesn't reject unless the network connection fails. Otherwise the response resolves, and the HTTP status code must be checked in `then`. Here's an ES6/ES2015 fetch wrapper to handle unsuccessful...

> Yes if the node modules don't use node. I think this means things like `require` aren't available, because it's a higher-level Node API, not V8. So, many modules will...

@sb-bilal-dev-ev as I [commented](https://github.com/rogchap/v8go/issues/148#issuecomment-903031001) above, `require` is a Node API. It's not part of V8. It won't work in this context.

I'm currently testing this change with various engines and large batches.

Tested this new `pio batchpredict` with all three model types: * ✅ custom PersistentModel (ALS Recommendation) * ✅ built-in, default model serialization (Classification) * ✅ null model (Universal Recommender) This...

BTW, I found performance for a large, 250K query batch running on a single multi-core machine is equivalent to the previous Spark RDD-based performance.