[Docs] Should we need a feature comparison table for the various deployments?

[Rigo-m]

For example: I was taking a look at nitro tasks feature, and I was wondering:

• Do they work in every preset?
• Do they have limitations in some environments? (E.g: netlify functions have a cap on runtime of 10 seconds if you use regular functions, do tasks use background functions? Do they not? Does cron tasks works everywhere or only in some environments? This is not clear)

This is just an example, I was wondering if it should be correct to implement and maintain a documentation part with every quirks you might find for each feature in the various providers.

[pi0]

Making a (verified) overview of deployment runtime limits like max exec time might be a good idea, and we can list them in each.

For features, it might be tricky because runtime limits can be extended using 3rd party integrations or agnostic solution (like invoking tasks outside of netlify functions or a feature emulated) so i prefer if we avoid mentioning vendor names in features sections.

[Rigo-m]

I understand what you're saying!

So picture these two scenarios:

1. I am building a project, I am using X features (store, db, cache, tasks), everything is happily working in local development. I am confident everything will be working (minus some minor inconveniences), I then discover that on netlify Y,Z features doesnt work, and I wasn't warned anywhere that they dont
2. I have built a project, everything is working great thanks to nitro on X platform. I am confident that moving to Y platform will be easy since I have worked within nitro limits and boundaries (didn't use any X platform native technologies and packages, I have only use nitro default API), I then move to Y playform and I discover that some features are not supported there

So, provided that it is tricky, as you wisely said, to track runtime limits for features, how can we try (somehow, best effort) to avoid these 2 scenarios (or make them be less likely)?

[pi0]

Built-in agnostic features we introduce to the core should work, regardless of the deployment platform.

DB, Tasks, ALS, and WS are still marked as experimental because they can't match this promise. We should fix the limits.

Execution limit looks like a fairly good thing to start by documenting in deployment pages.

Do you know other "runtime" limits that might be worth mentioning?