-
Notifications
You must be signed in to change notification settings - Fork 7.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Operator Window #1123
Operator Window #1123
Conversation
RxJava-pull-requests #1035 ABORTED |
Not sure why it aborts, locally, it fails on the clojure test
|
Perhaps then it needs to use |
That can be arranged, but the current |
It is odd that it throws a This whole type is odd, because we are bridging a time boundary and avoiding the use of a multicast subject on a hot observable that we have created from a potentially cold observable. We do not add multicast anywhere unless the user asks for it, and probably shouldn't here either. I'm not sure the overhead of Do you think we should support multicast on |
In PR #1138, I've fixed BufferUntilSubscriber to behave as a simple PublishSubject for the subsequent subscribers. It works and most users compose operations that subscribe once, but it is not a least surprising behavior. I'd add a replayAll and replayOnce operators to GS so users have to explicitly request that behavior. Both would immediately subscribe to the group and route everything through ReplaySubject or BufferUntilSubscriber, which I would rather rename to ReplayOnceSubject. |
More clarity on the multicast issue ... if we were to start allowing multicast on a If someone wants to multicast they should have to explicitly ask and deal with the |
No, that's not how it should work. I specifically implemented It should not do multicast. It is single-purposed for hopping the time gap, and that's it. |
Operator Window
Issue #1060
Significant changes have been implemented in the new Window operator's behavior:
Observable.from
. This is not how the Rx.NET version behaves where once a window is open, values are delivered as they appear from source. This has the same effects asgroupBy
where Observable windows not subscribed immediately will not deliver all values. Probably this is why Clojure testtest-partition-all
fails, but I don't know how to fix it there.This rewrite needs more meticulous review due to the changes and their concurrency effects.