Have you thought of a devtool for derivable?
vis for atoms + derivation trees
updates
And I have considered devtools for Derivable but I'm not sure the kind of visual graph thing you described would be useful. I've thought about some browser devtools for a higher-level app framework I've been designing, but that relies on a particular kind of data store which doesn't exist yet.
then some highlights when values change
then some API for annotations in model (derivable) so one could scale devtools to any of the arch their app uses (flux, ...)
Interesting? Sure. Useful? I'm not convinced. How would it improve DX? I can't see it helping to grok a codebase, since the code is already declarative. One of the great benefits of derivables is that you only need to know about immediate parents in order to grok what a particular derivable is all about, and the immediate parents are right there in the source code.
I think seeing what atoms you have in app is useful because this is what your app is about, full picture. Guess there will single (usually) atom with app state, several with component local state (think React app), then atoms which bridges external state (browser history) — could be useful to have an API to tag atoms somehow (component local, app, history.
Then see what changes when you interact with the app.
Also seeing what reactors are active (could be also made with tagging). Like what components are subscribed to what and when they are re-rendered.
Sorry for mind dumping, I'll write a proposal maybe in a gist sometime.
react-derivable and in-house
Like 1) does the function passed to
.derive() method allows to deref other derivable? Will they be tracked then? 2) less API surface 3) consistent pattern for accessing derivable value d.get() vs. d.derive(v => ...)
My first reaction is that this would be a bad idea because .derive has all kinds of syntactic benefits that I personally use all the time. However, you raise valid points. And I like the idea of keeping API surface area minimal even at the cost of syntactic verbosity. I'll definitely think about this.
For debug/devtool purposes.
Looking at https://www.npmjs.com/package/mobx-logger now
@ds300 may I ask for a review a little bit later?
deriveInWorker<T, R>(source: Derivation<T>, derive: (value: T) => R): Derivation<?R>
Basically it returns
null if result isn't ready yet
My naive impl was to setup a reaction on
source and submit task to a worker and put result eventually into another atom I return from deriveInWorker. Then (important) I want not to do any work if there's no reactions active on the derivation I return.
But I'm not sure it is possible now to do that as I don't have access to derivation lifecycle
wrote a little intro to derivable: https://medium.com/@andreypopp/learn-derivable-in-two-minutes-1ea82b5654db#.bhs8h79vf
plan to make a series of small blog posts, each focused in a single topic
cc @ds300 ^^
https://medium.com/@andreypopp/4146107ed247 3rd (not yet published, private link)
