Conflate Operator

[mrudangit]

RxJava should have conflate operator. I see we have sample, debounce etc but actually they are not equivalent to conflate .

Conflate operator should behave as Sample if updates are faster than the given interval. But if updates stops and new update comes after a long period it should push immediately. Instead of waiting for sampling period.

[JakeWharton]

Is the behavior your after equivalent to zip(stream, interval())? I too
want such an operator and I was about to file an issue tomorrow. It finally
hit my three separate use cases bar for wanting it to be first party.

On Tue, Nov 15, 2016, 7:06 PM MajorMud notifications@github.com wrote:

RxJava should have conflate operator. I see we have sample, debounce etc
but actually they are not equivalent to conflate .

Conflate operator should behave as Sample if updates are faster than the
given interval. But if updates stops and new update comes after a long
period it should push immediately. Instead of waiting for sampling period.

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#4856, or mute the thread
https://github.com/notifications/unsubscribe-auth/AAEEEd0LKzKaRnG1inDQ8dVB6VichZr9ks5q-nNQgaJpZM4KzS88
.

[mrudangit]

Not sure how would you achieve with zip.

sample at timeout out interval if source is emitting too fast but if source is slower than the timeout interval emit right away.

O.O.O.O.O...............O
--->-->-->-->-->-->-->-->
----X-------X-------------X

[abersnaze]

Is throttleFirst closer to what you are looking for?

[JakeWharton]

throttleFirst drops events. I believe this doesn't want that. It's like "at
most" every X interval.

On Tue, Nov 15, 2016, 8:25 PM George Campbell notifications@github.com
wrote:

Is throttleFirst closer to what you are looking for?

https://camo.githubusercontent.com/fe3f3d248d4933e30866c27188277684b24cbef8/68747470733a2f2f7261772e6769746875622e636f6d2f77696b692f5265616374697665582f52784a6176612f696d616765732f72782d6f70657261746f72732f7468726f74746c6546697273742e706e67

—
You are receiving this because you commented.

Reply to this email directly, view it on GitHub
#4856 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AAEEEQHZBk0SiHGv1w-cS-l-XTh92gBaks5q-oWvgaJpZM4KzS88
.

[mrudangit]

throttleLast/Sample is what i want if source is publishing too fast but then publish the item right away don't wait for the sample timeout if it was published slower than then timeout interval.

took stab at this

https://github.com/mrudangit/HelloRxJava/blob/master/src/main/java/com/solutionarchitects/Conflation.java

[abersnaze]

How about this?

[JakeWharton]

For my case I would need green before the blues. If that's not conflate
I'll file a separate issue.

On Tue, Nov 15, 2016, 8:43 PM George Campbell notifications@github.com
wrote:

How about this?
[image: image]
https://cloud.githubusercontent.com/assets/406038/20335040/e0c40332-ab73-11e6-9238-17bd9ff8cbbc.png

—
You are receiving this because you commented.

Reply to this email directly, view it on GitHub
#4856 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AAEEEeugJuqKqFvD7OH2pRXgKPqcpkyfks5q-ooGgaJpZM4KzS88
.

[abersnaze]

I was being lazy about the timing of the end of yellow's timeout and cyan's arrival. I was mostly trying to highlight that this is throttleFirstAndLastButNotLastIfItIsTheSameAsTheFirst

• red is not emitted twice
• blue is emitted at the end of cyan's interval

[mrudangit]

Red should output at the end of first period, Green at the end of 2nd period, dark blue at the end of 3rd period , not at the start of period as shown. How to generate diagram. probably can explain that way better

[abersnaze]

Was I misinterpreting this statement?

publish the item right away don't wait for the sample timeout

That seems to conflict with

Red should output at the end of first period

The diagram source file is an OmniGraffle in the wiki repo https://github.com/ReactiveX/RxJava.wiki.git the file is images/rx-operators.graffle

[mrudangit]

[akarnokd]

If I understand correctly, you need a throttleFirst but instead of dropping the incoming values in the period, you'd want to keep the latest and emit it immediately once the period ends, starting a new period. What should happen if there is an active period plus a latest element is waiting and the main source completes?

[mrudangit]

emit the last one emitted before onComplete
yes want to throttle first
but if there is inactivity more than the throttle period
when source emits emit right away bcos we already waited more than the throttle periods

so in my diagram yellow is emitted right away even though it falls between the throttle period because it wait one full throttle period before

[akarnokd]

I think this operator should go into rxjava-extras/RxJava2Extensions as it seems to address the needs of a small set of developers only and possibly introduce confusion with throttleFirst. I'm sure @davidmoten and I will accept such operator if contributed.