What is the benefit of using stream vs just using the "convenience" .fresh or .get? about store

[eyalgu]

There are 2 main benefits to stream over a single value:

1. As you say, you can get values from disk/cache/fetcher, which is useful if you want to first display a cached value while waiting for the fetcher to get a fresh value (this is the main use case covered by the fresh extension function).
2. Stream allows you to continue receiving updates to the data as it changes. If you use the data coming out of store to show a UI for example, by using stream you'll be able to always show the newest data. New data will be emitted if changes were made to the sourceOfTruth or if another request caused store to refetch the item you are streaming.

For your 2nd question:
Every call to stream (or fresh) will cause a new cold flow to be created (which will not cause any coroutine or other non-builder action to be performed). The action starts when you collect the flow (the equivalent of subscribe if you're coming from Rx world). Store, under the hood will coalesce all the fresh calls into a single fetcher call. Generally, store's stream should never be stopped by store, only by the user (though maybe there are some error conditions that we'll cause it to stop? if so we might want to fix those to make it clear store will emit until you're the collection is stopped by the user).

So what I'm hearing is that we can call StoreRequest.fresh() as much times as we want and when the collector is active it will start a new flow but we don't have to worry about deleting previous flow objects that were created from function calls.

How does it work if we open both a fresh and cache stream? Calling fresh will also update the cache, which means that cache streams will also receive a new emission, which could introduce possible duplication of the displaying of the data to the UI.

[eyalgu]

So what I'm hearing is that we can call StoreRequest.fresh() as much times as we want and when the collector is active it will start a new flow but we don't have to worry about deleting previous flow objects that were created from function calls.

Yes. You can collect for as long as you care about the results. The collection will be automatically cancelled when the collecting coroutine (or coroutine scope) is completed/cancelled.

How does it work if we open both a fresh and cache stream? Calling fresh will also update the cache, which means that cache streams will also receive a new emission, which could introduce possible duplication of the displaying of the data to the UI.

Fresh and cached refer to the initial results - is it ok to get a cached value? Should we avoid calling the fetcher? Etc. That does not affect subsequent update. If you asked to stream the data, that mean you will get all the updated that store knows about to the data your collecting until you stop the collection