You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@atlas.apache.org by "Nigel Jones (JIRA)" <ji...@apache.org> on 2016/04/01 12:12:25 UTC

[jira] [Commented] (ATLAS-518) Tinkerpop3 - graph store abstraction

    [ https://issues.apache.org/jira/browse/ATLAS-518?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15221497#comment-15221497 ] 

Nigel Jones commented on ATLAS-518:
-----------------------------------

See also proposal for titan abstraction in ATLAS-610

> Tinkerpop3 - graph store abstraction
> ------------------------------------
>
>                 Key: ATLAS-518
>                 URL: https://issues.apache.org/jira/browse/ATLAS-518
>             Project: Atlas
>          Issue Type: Wish
>    Affects Versions: trunk
>            Reporter: Nigel Jones
>
> Atlas is currently based on tinkerpop2 & titan 0.5.4. Work is needed to bring it up to the current tinkerpop3 level which would be covered by any titan upgrade as mentioned in ATLAS-517.
> However one role of the very active tinkerpop project is to provide a layer of abstraction between a graph data store, and applications (such as Atlas)
> (See http://tinkerpop.apache.org/docs/3.1.1-incubating/reference/ )
> By adopting tinkerpop interfaces where possible we can facilitate integration/interop with other graph stores and offer much more flexibility in implementation as appropriate.
> Tinkerpop (or gremlin as the term is now used for the whole stack) isn't complete -- for example creation of indices today relies on defining the index in the underlying implementation (in our case titan), but we also have the opportunity to work together with the tinkerpop team to reduce these interdependencies moving forward
> From a scan of the code Atlas today uses titan APIs and it would require some effort to migrate. This JIRA is being opened to get discussion on whether this direction makes sense, to understand the pros/cons of such an approach, and to evaluate the impact/feasibility of making such a change.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)