Meteor: ReactiveVar vs ReactiveDict

Question

I understand that the difference between using ReactiveVar and Session variables is, at the basic level, a matter of local vs global variables. ReactiveDict is like a local Session object.

But why use ReactiveDict at all if you're already using ReactiveVar? ReactiveDict doesn't seem to have any advantages over ReactiveVar

Answer 1

Here are the main differences between ReactiveDict (RD) and ReactiveVar (RV):

1. As its name implies, RD offers dictionary-like semantics: set takes a key/value pair and it has delete and clear methods. The bottom line is it lets you cluster reactive data into a single variable. Frankly, this isn't all that interesting in most use cases, but keep reading.
2. RD has an all method which returns all of the key/value pairs and registers a reactive dependency on each of them. The main benefit is that if you have a cluster of related reactive data you can fire an autorun whenever any part of the dictionary changes. To do this with a set of RVs, you'd need to call get on each of them.
3. If you pass a name into the RD constructor, it will survive a hot code push. In fact, Session is actually just an RD.

In practice, I've found a few use cases for RDs, but RVs are what you'll likely choose in most situations.

Recommended reading: the ReactiveDict source code

Answer 2

"... a general-purpose reactive datatype for use with tracker."

"If you provide a name to its constructor, its contents will be saved across Hot Code Push client code updates."

source: https://atmospherejs.com/meteor/reactive-dict