incubator-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Marko Rodriguez <okramma...@gmail.com>
Subject [Proposal] TinkerPop: A Graph Computing Framework
Date Wed, 17 Dec 2014 19:09:48 GMT
Hello,

My name is Marko A. Rodriguez and am a co-founder of TinkerPop (http://tinkerpop.com). There
has been positive pressure on us (both internally and externally) to move TinkerPop to The
Apache Foundation. This email contains our proposal and I, on behalf of TinkerPop, thank you
for spending your time reading it.

Take care,
Marko.

http://markorodriguez.com

------------------------------------




A. Abstract

TinkerPop is a graph computing framework written in Java -- http://tinkerpop.com. TinkerPop
started in 2009 and is currently in the milestone series of 3.0.0.  Graph system vendors such
as in-memory graph libraries, OLTP graph databases, and OLAP graph processors can provide
a TinkerPop implementation (http://markorodriguez.com/2013/01/09/on-graph-computing/). Once
the core TinkerPop interfaces are implemented, the underlying graph system can be queried
using the graph traversal language Gremlin. TinkerPop was designed to be used at any scale
and as such, can be leveraged by both single-server and multi-machine compute clusters. For
many, TinkerPop is seen as the JDBC of the graph computing community (http://en.wikipedia.org/wiki/Java_Database_Connectivity).

B. Proposal

Since its inception in 2009, TinkerPop has been open source, free to use. Initially the license
was BSD, but as of TinkerPop3, the license is Apache2. The TinkerPop team is composed of developers
as well as representatives from numerous graph system vendors (http://www.tinkerpop.com/docs/3.0.0-SNAPSHOT/#tinkerpop-contributors).
TinkerPop has done its best to remain vendor agnostic and works closely with vendors to ensure
that the constructs within TinkerPop are able to accommodate the requirements of the underlying
graph system. To date, over 12 graph system vendors provide TinkerPop implementations. We
believe that by joining The Apache Foundation, our vendors, users, and contributors will feel
more comfortable in terms of legal protected, in terms of wider-adoption, and in terms of
project stability.

C. Background

TinkerPop was founded in 2009 and has had steady, active development since. Over the years,
the Gremlin query language within TinkerPop has been adopted by various JVM languages and
as such, there exists Gremlin-Groovy, Gremlin-Scala, Gremlin-Clojure, Gremlin-JavaScript,
and the like. In many ways, Gremlin is seen as a traversal language style that can be readily
adopted by developers --- on and off the JVM. TinkerPop is not bound to the JVM in that developers
wishing to interact with a TinkerPop-enabled graph system can leverage Gremlin Server which
provides "over the wire" communication as well as the entry point for non-JVM language bindings.
TinkerPop is being used is production graph-based applications around the world and is only
getting better with age.

D. Rationale

The graph computing space has grown over the years to encompass numerous graph database and
graph processing systems. TinkerPop was created as a unifying framework for interoperability,
language standardization, and data model standardization.  This framework makes it simple
to "plug and play" the back-end graph implementation without affecting the developer's code.
This is analogous to the way in which JDBC allows users to swap relational databases while
keeping the same programming interface. TinkerPop also brings together OLTP systems (graph
databases) and OLAP systems (graph processors) by providing a single query language, Gremlin,
for executing graph algorithms transparently over either type of system.  Finally, TinkerPop
unifies single-machine systems and distributed systems, presenting an identical user experience
within the boundaries of the computational space and time constraints of the underlying graph
system.

E. Initial Goals

The goal of this proposal is to migrate TinkerPop to The Apache Foundation to 1.) get more
exposure to the project and 2.) legally protect the developers and contributors of TinkerPop.


F. Current Status

TinkerPop is planning to release TinkerPop 3.0.0.GA early 2015 -- January or February. We
would like to release TinkerPop 3.0.0.GA under The Apache Foundation.

G. Meritocracy

Anyone is welcome to join TinkerPop as a contributor. If an individual provides successful
code updates, documentation updates, etc. then they are asked to join TinkerPop contributors.
Once inside TinkerPop, they are able to voice their opinions/thoughts on the direction of
the project. Moreover, TinkerPop actively seeks vendors who leverage TinkerPop in their offering
to place a representative on TinkerPop contributors who can speak on behalf of their organization
as it relates to the requirements of their graph system. Every year or so, we list inactive
members and request their permission to purge them from TinkerPop contributors. In this way,
we try and keep the TinkerPop contributors to the most active, participating members.

H. Community

There are 25 TinkerPop contributors which, of those, 12 are representatives from the various
graph system vendors in the market today. The TinkerPop mailing list has over 1500+ subscribers
with a flow of approximately 10 emails a day  
(https://groups.google.com/forum/#!forum/gremlin-users). 

I. Core developers

The primary developers are Marko A. Rodriguez (http://markorodriguez.com) and Stephen Mallette
(http://stephen.genoprime.com/). Over the years, numerous individuals have come and gone from
TinkerPop. Other significant contributors include Peter Neubauer, Joshua Shinavier, James
Thornton, Matthias Bröcheler, Daniel Kuppitz, Bryn Cooke, and others. Note that there are
numerous individuals that participate in the TinkerPop ecosystem by providing language bindings
and system implementations. While not contributing to the core codebase per se, they support
TinkerPop via their software and lecture outreach.

J. Alignment 

TinkerPop is licensed Apache2. The OLAP reference implementation that distributes with TinkerPop
is based on Hadoop (http://hadoop.apache.org) and Giraph (http://giraph.apache.org). There
has been a push from vendors and commercial organizations interested in ensuring TinkerPop
has the legal support of a software foundation. We believe The Apache Foundation to be a great
home for TinkerPop.

K. Known Risks & Abandoned projects

TinkerPop is currently at TinkerPop 3.0.0.M6. By joining The Apache Foundation, there is a
significant amount of code and time that has passed since TinkerPop's inception. However,
we have done our best to secure CLAs from contributors and have never gotten into a legal
or political issue.

L. Inexperience with Open Source

The primary developers of TinkerPop have been providing TinkerPop source and releases open
source and in full public view since 2009 via both our GitHub repository and our public mailing
lists. We pride ourselves in professional software, documentation, and aesthetics. With its
integration to numerous graph systems in the market, TinkerPop is truly a masterpiece in a
open source technology --- working with both open source and commercial entities to ensure
a solid, respected project.

M. Homogeneous Developers

The contributors to TinkerPop run the full gamut of software engineering expertise. From low-level
data serialization to high-level server interactions, TinkerPop has contributors able to work
at the requisite level. The codebase is complex as it deals with numerous area of software
engineering including: distributed computing, open standards, language bindings, query language
compilers, OLTP databases, OLAP processors, etc. Over the years, we have gathered a significant
level of talent and as a whole, have all grown as engineers.

N. Reliance on Salaried Developers

TinkerPop is open source and no individual is paid for their contributions except for our
logo designer. All the logos and documentation graphics are released Apache2.

O. Relationships with Other Apache Projects

TinkerPop currently leverages Hadoop, Giraph, and in the near future, will integrate with
Spark. Internally, the project leverages Maven3, Apache Configuration, and Apache Commons.

P. An Excessive Fascination with the Apache Brand

TinkerPop is primarily interested in The Apache Foundation for the legal support and protection
it can offer our developers and users. Numerous large software organizations have urged TinkerPop
to join a foundation so that they may more easily provide contributors from their organization
and moreover, maintain confidence in building products atop TinkerPop. With that said, TinkerPop
does respect The Apache Foundation brand and believes that by joining The Apache Foundation,
we will further boost the recognition and prestige of our product. Finally, note that a TinkerPop
book is slated for 2015 (http://www.tinkerpopbook.com/). Combining The Apache Foundation and
a long awaited book should ensure a new uptake of contributors, users, and promoters.

Q. Documentation

TinkerPop takes pride in its documentation:
	AsciiDoc: http://www.tinkerpop.com/docs/3.0.0-SNAPSHOT/
	JavaDoc (core): http://www.tinkerpop.com/javadocs/3.0.0-SNAPSHOT/core/
	JavaDoc (full): http://www.tinkerpop.com/javadocs/3.0.0-SNAPSHOT/full/

R. Initial Source

TinkerPop is currently hosted on GitHub.
	https://github.com/tinkerpop/tinkerpop3

S. Source & Intellectual Property Submission Plan

There are no known claims to violations of intellectual property.

T. External Dependencies 

TinkerPop relies on open source and commercial vendors. We maintain representatives from the
various graph system projects on TinkerPop contributors. These vendors, in turn, provide TinkerPop
adaptors to their technology and in fact, some of them provide TinkerPop as the sole interface
to their system.

U. Cryptography

None.

V. Required Resources

The only resource we see moving forward and if Apache wishes, is to fund our logo designer
when we need a logo. We currently spend about $500 a year on logos. Please look through our
documentation to see how much we love giving character to our work. If this is not something
that Apache can do, then we will continue funding the creation of our artwork as we have done
in the past.

W. Mailing Lists

We currently have 2 mailing lists:
	Gremlin-Users (public user forum): https://groups.google.com/forum/#!forum/gremlin-users
	TinkerPop-Contributors (private contributors forum): <private>

X. Subversion Directory

TinkerPop uses GitHub for its source control.

Y. Git Repository

TinkerPop's GitHub account is: https://github.com/tinkerpop/tinkerpop3

Z. Issue Tracking

TinkerPop uses GitHub's issue tracker for tracking issues: https://github.com/tinkerpop/tinkerpop3/issues

AA. Initial Committers

TinkerPop's contributors are listed here and all those that provide source code commits have
signed a CLA.
	http://www.tinkerpop.com/docs/3.0.0.M6/#_graph_developers

BB. Affiliations 

None.

CC. Sponsors

IBM (Kelvin Lawrence and Todd Moore) were the primary proponents of pushing TinkerPop to join
The Apache Foundation. The vendors that support TinkerPop are listed here: http://www.tinkerpop.com/docs/3.0.0.M6/#_graph_vendors




Mime
View raw message