You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@edgent.apache.org by "Dale LaBossiere (JIRA)" <ji...@apache.org> on 2017/03/02 15:16:45 UTC
[jira] [Updated] (EDGENT-27) Additional documentation for
publish-subscribe.
[ https://issues.apache.org/jira/browse/EDGENT-27?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Dale LaBossiere updated EDGENT-27:
----------------------------------
Component/s: Documentation
Connectors
> Additional documentation for publish-subscribe.
> -----------------------------------------------
>
> Key: EDGENT-27
> URL: https://issues.apache.org/jira/browse/EDGENT-27
> Project: Edgent
> Issue Type: Test
> Components: Connectors, Documentation
> Reporter: Daniel John Debrunner
> Assignee: Daniel John Debrunner
> Priority: Minor
>
> From: https://github.com/quarks-edge/quarks/pull/98
> generally +1 on all of this!
> I think more doc is needed to inform users about the characteristics of the current impl:
> Clarify there's no buffering/memory for tuples added to a topic: a published tuple is delivered only to current subscribers. The current wording focuses only on the no-subscribers case.
> Must manually add a PublishSubscribeService to a provider instance to use this service. An example code fragment in the class would be beneficial.
> The ProviderPubSub impl only works between topologies from the same provider instance
> A publishing stream is isolated from its subscribers, and the subscribers from each other, via an implicit PlumbingStreams.isolate() on each subscribing stream. Users may want/need to manually insert a PlumbingStreams.pressureReliever() on subscriber streams to avoid excessive memory resource use from subscribers that can't keep up with the tuple production rate.
> A sample demonstrating pub/sub use would be beneficial.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)