[
https://issues.apache.org/jira/browse/TINKERPOP-1174?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15931174#comment-15931174
]
stephen mallette commented on TINKERPOP-1174:
---------------------------------------------
Another technical reason to implement this is that we will see improved cache
hits in Gremlin Server. For example, if you do:
{code}
g.addV("person").property("name",name,"age",age,"optional",opt)
g.addV("person").property("name",name,"age",age)
{code}
you'l get two different compilations, but if you do:
{code}
g.addV("person").property(props)
{code}
you get one compilation and have the ability to vary the properties.
This approach also has the possible benefit of making {{Bytecode}}
significantly smaller and traversal building far less expensive if the {{Map}}
of arguments was special cased to avoid use of {{Parameters}} and {{Bytecode}}
skipped adding "property" steps to its instruction list.
> addVertex(Map properties) method
> --------------------------------
>
> Key: TINKERPOP-1174
> URL: https://issues.apache.org/jira/browse/TINKERPOP-1174
> Project: TinkerPop
> Issue Type: Improvement
> Components: structure
> Affects Versions: 3.1.1-incubating
> Reporter: Matthias Broecheler
> Assignee: stephen mallette
>
> which overloads the other {{addVertex(Object... properties)}} method with a
> map based version (of key-value pairs).
> Having this alternative to addVertex is nice because:
> 1) It allows you to submit a map when adding vertices on remote graphs via
> drivers. Right now, you have to convert the properties to an array which is
> awkward and unnecessary
> 2) It would work well with groovy's map based syntax so one can write:
> {{g.addVertex( key1 : "value1", key2: "value2")}}
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
