Calculate upload rate

[gilest]

Continuation of work by @RobbieTheWagner in #1015

Methodology

• Record rate data against each file as upload progress events occur
• Calculate weighted moving average of rate per-file as File.rate property
• Sum all uploading files as Queue.rate property

All rates are in bytes/ms.

Demo

Recent improvements to the demo page make this much easier to validate and test.

Screen.Recording.2023-12-01.at.3.07.05.PM.mov

[RobbieTheWagner]

@gilest this is looking great! Quick question, should we be throttling this calculation instead of updating it on every event?

[gilest]

Would it be more performant to use a TrackedArray?

[RobbieTheWagner]

I'm not sure. I say leave it unless we have problems.

[gilest]

Work done in plain functions for unit-testing purposes.

[gilest]

@gilest this is looking great! Quick question, should we be throttling this calculation instead of updating it on every event?

Interesting question 🤔

So each time a progress event occurs we update the following properties:

• file.bytesWhenProgressLastUpdated
• file.timestampWhenProgressLastUpdated
• file.rates

Beyond that, no work is done. We'll only do a rate calculation if one of the rate getters is accessed. That's the beauty of lazy evaluation and derived state.

Obviously if a rate property is bound to a template or other reactive context then the rate calculation will run every time file.rates updates, but it only runs on 30 recorded rates per upload-in-progess. I doubt it's very computationally intensive.

So I don't think we need to do any throttling.

It's also worth noting that the demo simulation sends 10 progress events per second. Probably it's a bit less often in real-world HTTP uploads.

[gilest]

Profiling with 6x CPU slowdown the estimatedRate function takes ~1ms at most

Comparison to date-fns intervalToDuration in the same period

[RobbieTheWagner]

@gilest yeah seems performant enough and we can always kick the tires a bit and see if we need to tweak it later.

[RobbieTheWagner]

This looks great to me. Thanks for taking it over the finish line!