Performance observability

As you start your TypeScript journey (or even before!) I suggest adding some observability across all the major pipeline steps. For example, if I have 100 pages being generated but 5 of them take 95% of time to generate, we should be able to see that when the pages are being tracked. For large sites some parts of site take much longer to generate so the developer should be responsible for caching those pages but she’d need to be told which ones they are.

Something like this?

🔥 Deno prepare (900ms)
🔥 /images /images (copy: 25ms)
🔥 /index.html /index.tmpl.ts (load: 1025ms, render: 255ms, process: 190ms, etc: ??)
🔥 /styles.css /styles.css (??)

🍾 Site built into ./_site (1255ms)

It would be good to build OpenTelemetry and OpenMetrics (Prometheus) into Lume so that large sites can fine tune performance through observability.

[oscarotero]

Ok, but I don’t want to include dependencies or big developments for this in the Lume core. I prefer a basic solution that can be extended with plugins. The print and save actions are here: https://github.com/lumeland/lume/blob/master/metrics.js#L39 so it would be easy to create a plugin that can configure or replace this default solution with something more powerful or adapted to specific needs. A plugin example could be:

import Metrics from "lume/metrics.js";

class MyCustomMetrics extends Metrics {
  print() {
    // Custom print
  }
  async save(file) {
    // Custom save
  }
}

export default function () {
  return (site) => site.metrics = new MyCustomMetrics();
}

[oscarotero]

Ok, but note that on importing the dependencies statically, lume upgrade [--dev] won’t work because you are stuck to a fixed version of lume.