[
https://issues.apache.org/jira/browse/TINKERPOP3-921?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Marko A. Rodriguez closed TINKERPOP3-921.
-----------------------------------------
Resolution: Won't Fix
Will write up a new ticket around a {{Bulk}} class.
> A backwards compatible solution to the bulk/sack situation
> ----------------------------------------------------------
>
> Key: TINKERPOP3-921
> URL: https://issues.apache.org/jira/browse/TINKERPOP3-921
> Project: TinkerPop 3
> Issue Type: Improvement
> Components: process
> Affects Versions: 3.0.2-incubating
> Reporter: Marko A. Rodriguez
> Assignee: Marko A. Rodriguez
>
> This ticket is more a proposal/brainstorm. I was thinking we should abandon
> the concept of sack and bulk and simply have "Gremlin guts." :). Bare with
> me, its backwards compatible... so its not as crazy as it sounds.
> We should have an interface (where "Guts" is just a place holder for now):
> {code}
> public interface Guts<T extends Guts> {
> public T merge(T other);
> public T split();
> }
> {code}
> What most people do is NO sack and a bulk that incrs a long with merge being
> bulk sums. This would be the "Gremlin guts" of:
> {code}
> public BulkGuts implements Guts<BulkGuts> {
> public BulkGuts merge(BulkGuts other) {
> this.bulk = this.bulk + other.bulk;
> }
> public BulkGuts split() {
> return new BulkGuts(this.bulk);
> }
> }
> {code}
> Now, whenever a Gremlin splits/mergers, its guts are respectively split and
> merged. You want sacks without bulk, well, you have a Gut that does real
> numbers... no such thing as bulk.
> This makes it so the concept of "bulk" and "sack" are NOT part of
> {{Traverser}}, but are accessible via:
> {code}
> public interface Traverser<T extends Guts> {
> T getGuts();
> }
> {code}
> .... This isn't fully thought out, but by 3.2.0, when we release the full
> virtual machine specification, this may come into play.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
