Batched updates

[anandthakker]

Depending on the store/action design of an app, batching updates can be a pretty useful/important perf optimization. @yoshuawuyts what are you thoughts on the best choo-like way to handle this? Would you be interested in adding a hook that would make this possible in some app-wide way (like maybe a way to provide 'middleware' function through which updated states were passed before being emitted to the view-rendering)? Or would you prefer for people to handle this by simply throttling individual view functions?

[yoshuawuyts]

Ooh, yeah I've been considering this; combined with documenting best practices on thunking views and caching routes (yoshuawuyts/sheet-router#17) it would def be worthwhile to put an upper cap on flushing updates.

In the past for virtual-dom applications I used yoshuawuyts/virtual-raf to flush updates at 60fps; it could def be worthwhile exploring this for choo too - especially the interaction with webworker rendering would become interesting (max-mapper/yo-yo#22).

I think this should happen after the 3.0 release though; as it's a minor tweak without practical API changes. Does this answer what you meant? Cheers!

[yoshuawuyts]

Created https://github.com/yoshuawuyts/nanoraf yest; might add this. Talked about this on IRC, and considered not doing batched updates; but realized yesterday that sometimes something like a higher order effect might trigger a whole slew of updates in rapid succession because stores are separated - there'd be no way of throttling that so we should prob def do this. If not today, then soon ✨

[yoshuawuyts]

Created a new branch: https://github.com/yoshuawuyts/choo/tree/raf which uses nanoraf - it's based off the 3.0 branch and adds full batched updates.

This won't be part of 3.0 however as we probably need to test it a little, and I'd rather put out 3.0 before then - but it's non-breaking so can de make it 3.1 or smth ✨ Only difference that occurs is that older (<IE10) browsers will need a shim. I don't think that's bad enough to call a breaking change as it will realistically affect less than 1% of browsers out there, probably less for this lil choo thing we got here.

Anway, it can be played with now; keen to hear your thoughts on it.

[anandthakker]

@yoshuawuyts I haven't given it an actual test drive, but from a read of the code the approach looks great to me! I have a couple of thoughts but I'll post them as issues over in nanoraf.

[yoshuawuyts]

This was merged in 3.0.0 - seems to be working excellently 🎉 - thanks for suggesting it @anandthakker