I guess this is a workaround although it is kinda...

# announcements

[Andrew Gazelka]

I guess this is a workaround although it is kinda jank

[gildor]

Yes, suspending delegates are not available now at least because properties which do not support suspend modifier, at least for now
Do you have particular use case for this? I'm not really sure what you doing in your code and why do you need delegate there instead of simple assignment, just:
val a = aAsync.await(), without any delegation

[Andrew Gazelka]

alright so I am taking two `Flow`s of data and I am passing them into a HOF which will do calculations over time and want the latest value of the Flow (which will update over time) ... the reason for suspending at first is to wait until the first value from `Flow`s are returned

[gildor]

I mean that usually, if you want to do some async operation only once (as lazy works), you can just create async, save it to some field and than clients just call await() on the same instance, each of them will receive the same result, await will be run only once

[gildor]

Hmm, hard to say without example of code, not really sure that you can use Flow for such case, but maybe you could give a bit more context

[gildor]

What is HOF?

[gildor]

Just if you have 2 Flows and want somehow combine their data, shouldn't you use some zip or combineLatest operator, depending on case

[Andrew Gazelka]

HOF is a higher order function

[Andrew Gazelka]

also I am not zipping data

[Andrew Gazelka]

I am essentially turning the two flows into sources which can be polled from another long-running suspending function

[gildor]

Polled? Hmm, but it's cold stream, there is no polling of data, it's push based source

[gildor]

Looks that you need some Conflated broadcast channel to poll data

[gildor]

But hard to day without some sample

[Andrew Gazelka]

I think that makes sense, I am using a websocket to get prices of currencies

[Andrew Gazelka]

so initially it is a cold stream (new websocket created with initial function) but it should then be converted to a hot stream when referenced by functions which require polling

[Andrew Gazelka]

@gildor a ConflatedBroadcastChannel looks nice but I need a good way to suspend until the first value is received... any ideas?

[gildor]

Yes, use receive()

[Andrew Gazelka]

@gildor but receive() removes the last element... what if someone called receive() twice for one value? ...

[gildor]

But you cannot remove element from ConflatedBroadcastChannel

[gildor]

ConflatedBroadcastChannel.receive() you will just get current value

[Andrew Gazelka]

there is no receive() method in ConflatedBroadcastChannel

[Andrew Gazelka]

I mean I guess I could use Mutexes but aren't those generally seen as bad...

[gildor]

ConflatedBroadcastChannel.openSubscription().receive()

[gildor]

or even better just ConflatedBroadcastChannel.value

[Andrew Gazelka]

but I need to suspend until the first value is received

[Andrew Gazelka]

so ConflatedBroadcastChannel.value does not work

[gildor]

ah you right, I forgot that you want await for the first value, just ConflatedBroadcastChannel.openSubscription().receive() than

[Andrew Gazelka]

wouldn't opening a new subscription each time be resource intensive?

[Andrew Gazelka]

or it just seems weird... probably just me tho

[gildor]

or it just seems weird

It’s not weird, this stream of events, you have to open it, even if this stream is actually observable data

[Andrew Gazelka]

but you are opening it every time you need to poll a value

[gildor]

But careful, you probably also want to close subscription, so to make it a bit safer you can use flow instead:

ConflatedBroadcastChannel.asFlow().single() // or singleOrNull() if you don't want to throw exception

[gildor]

but you are opening it every time you need to poll a value (edited)

yes, but what do you expect? Or you await it and than you need some primitive, or you just get current value

[gildor]

I still not sure about your case, but it’s way how data streams work, not only reactive

[Andrew Gazelka]

my use case is I have two websockets for data (prices) which is essentially an event-driven and makes sense to use a Flow (even though normally hot data... but the web socket is created with a function so it is self-contained and is still cold) ... then I am using the two flows in another function which does period calculation given the current price... the Flow is used for updates in price.

[gildor]

period calculation given the current price

You shouldn’t use pull for this

[gildor]

or just wait for new data, or request this data by doing force update (just create a method for this)

[Andrew Gazelka]

I am getting the data from a websocket and I do not control the API

[gildor]

So than you don’t need any kind polling

[Andrew Gazelka]

however, once I get the data from the event-driven web socket I need to access it in a polling-like way

[gildor]

why?

[gildor]

if you want to cache use conflatedbroadcast channel + flow

[Andrew Gazelka]

the formula needs to request what the current last value of data is

[Andrew Gazelka]

yeah ^^^

[Andrew Gazelka]

using CBC and flow is fine

[gildor]

request what the current last value of data

but you said that you cannot force request last data you do not control API

[Andrew Gazelka]

I am using a web socket... then have a black box in my program which looks at data in my web socket and lets me request the last value

[gildor]

so what is problem with CBC, flow and combineLatest?

[Andrew Gazelka]

none I think...

[Andrew Gazelka]

which is what I was trying to say... don't think I was clear tho

[gildor]

ah, sorry

[gildor]

I just thought that we discussing how to do that

[gildor]

Did you see kotlinx.corotuines 1.3 release notes?
https://github.com/Kotlin/kotlinx.coroutines/releases

[gildor]

channelFlow is what you need to wrap webscoket

[gildor]

with bufferSize = Channel.CONFLATED param

[gildor]

than use combine latest for 2 flows to get updates when one flows is changed using combineLatest:
https://github.com/Kotlin/kotlinx.coroutines/issues/1193

[louiscad]

Also, #C1CFAFJSK 🙂