Does Firebase offload the computation for `NormalizedCollection` ? about firebase-util

[katowulf]

All joining is done client side. This doesn't improve the download/upload performance, although it is fairly performant about how it conducts joins/et al. Note that this optimizes writes as well as reads, pushing set()/update() ops back to the correct paths.

Cheers,
Kato

[gigablox]

Got ya, so the joins are done client side.

What about the data coming back? Is that combined into a single payload or is it more of a promise array type deal?

[katowulf]

Have you checked out any of the examples? They show the payload for each request.

[gigablox]

I guess the answer to my question was:

"This doesn't improve the download/upload performance, although it is fairly performant about how it conducts joins/et al."

Wondering if you guys are ever going to kick this up to the server in the future?

In my example if Jim has 500 older messages he want's to browse through they all need to be queried through a promise array.

I'm sure there's got to be a tipping point somewhere... any idea at what point the browser will start to struggle waiting for that payload to resolve and build itself?

[katowulf]

It's going to be highly subjective--really a product of the computer's computing capacity and memory. It seems like the best server-side solution would be to pre-compile the read data into a separate path. Some upcoming tools would assist with this.

The performant solution for large data would probably be a server-side component, such as a node.js script, which would monitor the data and compile the joined results for read. Then clients would only need to listen on the compiled path for the pre-joined and rendered data.