Diagnostics channels

[bjohansebas]

I’m opening this to discuss which modules are essential and make sense to add this API to, and where in the package they would go, this would be essential to improve the traceability of Express applications

• expressjs: Request and Response Functions
• router: Layer

Feel free to edit which packages need the integration.

[wesleytodd]

pillarjs/router#96
expressjs/express#6353

I feel like there are more open issues, but I couldn't remember any of them off hand.

Ah, this one is related but not exactly the same: expressjs/discussions#341

And I think we will need to look at body-parser, serve-static, send and probably some of the middleware packages.

[wesleytodd]

Looking for a champion on this work. Also may consider asking APM folks to come help us understand the needs and maybe remove some usage of debug.

[bjohansebas]

As I mentioned in the meeting, for now I'm only interested in adding it to the router package. However, before that, I need to finish some other pending work I have in that package, so it will take me some time to integrate it. If someone else wants to move forward with it in the meantime, that's totally fine with me.

[kjugi]

I was trying to get some more detailed overview on this topic. There are many different directions now. I will point everything in the end. As a starting point I can see this Draft PR pillarjs/router#96
This connects with @bjohansebas comment for implementing it in router package.

Based on the OP comment this PR needs to be reworked to be up-to-date. I can at least take a look on this. I'm planning to compare it with what's currently stable in the nodejs docs for diagnostics channels and compare with implementation from fastify. The second part with fastify is just for me to make sure I look on something done right.
After that I can let you know if I understand anything from it.

Except from that I can see there are other topics that fly around:

• considering OpenTelemetry. I'm not sure if this conflicts here with anything as they operate on different levels and probably different purpose (?) I feel this can be done as a separate activity
• replacing debug with any other package. Again, I can see this as another separate small PR. Also, what should be prepared except PR with replacement? Some benchmarks, tests, comparison with other solutions?
• implementations in different repositories than router. I think we can decide which one to qualify later and focus to deliver one draft implementation in router, right?