[
https://issues.apache.org/jira/browse/TINKERPOP-1497?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
stephen mallette closed TINKERPOP-1497.
---------------------------------------
Resolution: Won't Do
I randomly brought this one up with kuppitz while perusing old issues and he
mentioned that he wasn't able to recall a situation where we'd hit OME due to
barriers. I can't recall such a complaint either. If we haven't seen that yet,
it's probably safe to not worry about that issue until it arises and if it
does, we might question why that traversal wasn't using OLAP.
> Add BalancedBarrierStrategy to protect from OME
> -----------------------------------------------
>
> Key: TINKERPOP-1497
> URL: https://issues.apache.org/jira/browse/TINKERPOP-1497
> Project: TinkerPop
> Issue Type: Improvement
> Components: process
> Affects Versions: 3.2.2
> Reporter: Marko A. Rodriguez
> Priority: Major
>
> I'm not quite sure of the need for this feature, but will add it regardless
> so we can think on it more.
> There are various strategies that introduce "lazy barriers" into the
> traversal.
> * {{LazyBarrierStrategy}}
> * {{RepeatUnrollStrategy}}
> * {{PathRetractionStrategy}}
> Right now, the size of these barriers are hardcoded as "private static ints"
> inside of each respective strategy. The problem is that with all these
> barriers, we might get OME if all barriers are "full" (and we have many
> barriers).
> {{BalancedBarrierStrategy}} (a {{FinalizationStrategy}}) would have a "max
> total barrier size" and then would divide that barrier size amongst all the
> barriers (in global children) in the traversal. Perhaps the division can be
> biased by the use of filters, the depth in the traversal of the barrier, etc.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
