Presenter: Joshan V John Robert Dyer, Hoan Anh Nguyen, Hridesh Rajan & Tien N. Nguyen Iowa State University, USA Instructor: Christoph Csallner 1 Joshan

BOA: A LANGUAGE AND INFRASTRUCTURE FOR ANALYZING ULTRA-LARGE-SCALE

SOFTWARE REPOSITORIES

Presenter: Joshan V John

Robert Dyer, Hoan Anh Nguyen, Hridesh Rajan & Tien N. NguyenIowa State University, USA

Instructor: Christoph Csallner

1Joshan Valayil John | The University of Texas at Arlington | CSE 6324

Agenda

Motivation

Ultra-large-scale software repositories

Barriers to mining software repositories

Solution - Boa

Goals of Boa

Boa Architecture

EvaluationJoshan Valayil John | The University of Texas at Arlington | CSE 6324 2

Motivation

Big-3 software repositories known to have close to 1 million projects.

Contains a wealth of software and information about software.

Systematic extraction of relevant data from these repositories and their analysis for testing hypotheses is hard.

Boa, a domain-specific language and infrastructure, developed to ease testing ‘Mining Software Repository’ related hypotheses.

Joshan Valayil John | The University of Texas at Arlington | CSE 6324 3

Ultra-large-scale Software Repositories


Why analyze software repositories?

Curiosity

Identify patterns

Forecasting

Plan for better designs

Empirical Validation



Develop programming expertise to access version control system.

Establish infrastructure to store downloaded data from software repositories.

Joshan Valayil John | The University of Texas at Arlington | CSE 6324

Develop an application to access this local data. Improve scalability of analysis infrastructure to

process ultra-large-scale data.

6


Experiments are often irreproducible

Low reusability of experimental infrastructure

Lack of systematic curation leads to loss of

experimental data.

Building analysis infrastructure to process ultra-

large-scale data efficiently can be very hard.


Solution - Boa

Designed a domain specific language and

infrastructure to analyze ultra-large-scale

software repositories – Boa.


Goals of Boa

Easy to use

Better abstractions

Efficient & Scalable

Enhances reproducibility


A Research Question

Consider a program that answers:

“What are the churn rates for all Java projects

that use SVN?”


Solution in Java Full program over 70

lines of code. Uses JSON and SVN

libraries. Runs sequentially. Takes over 24 hours. Takes almost 3 hours

with data locally cached.

Can be parallelized, but very complex.


Solution in Boa


Simple program, 6 lines of code. Hides implementation specifics. Auto parallelization, results in 1 minute. Results can be easily reproduced by publishing these

small programs with the data sets used.12

Performance Results


Boa Architecture


Boa Architecture

Three main components

The Boa Language

Boa Compiler & Runtime

Supporting data infrastructure


The Boa Language

Domain-Specific Types

MapReduce Support

Quantifiers

User defined functions

Output Aggregators


Boa Language – Domain-Specific Types

Provides several domain-specific types which aid in abstracting mining software repository details (http://boa.cs.iastate.edu/docs/dsl-types.php)


Boa Language – MapReduce Support

Computations specified via two user-defined functions: Mapper – takes key-value pairs as input &

produces key-value pairs as output.

Reducer – Consumes the above output and aggregates data based on individual keys.


Boa Language – Quantifiers

Boa defines the quantifiers: exists foreach ifall


Boa Language – User-Defined Functions

Users can define their own mining algorithms

Facilitates code re-use.


Boa Language – Output aggregators


Output can be indexed

Output defined in terms of predefined data aggregators

21

Boa’s Supporting Infrastructure

Compiler & Runtime

Data Infrastructure

Web based interface


Boa’s Compiler & Runtime

Initial implementation was based upon the Sizzle compiler & framework

Sizzle is an open-source Java implementation of the Sawzall language.

Sizzle provides support for generating programs that run on the Hadoop open-source MapReduce framework.


Boa’s Data Infrastructure

Local cache of repository information. First Step – Locally replicate data.

Second Step – Run the caching translator to convert data into the framework required format.

Input (JSON file + SVN repositories) -> Output (Hadoop SequenceFile)


Boa’s Web based Interface

Submit programs.

Compile & run them on their clusters.

Each submission creates a job in the system.


Evaluation

Programs were executed on a Hadoop 1.0.3

install.

Cluster was not tuned for performance, except

for setting the maximum number of map tasks

for each compute node equal to the number of

cores on that node and increasing the VM heap

size.Joshan Valayil John | The University of Texas at Arlington | CSE 6324 26

Evaluation – Applicability

Research Question 1 – Does Boa help researchers analyze ultra-large-scale software repositories?

A set of 21 tasks in four different categories were examined. Programming Languages Project Management Legal Platform/Environment



Evaluation - Applicability


Evaluation - Scalability

Research Question 2 – Does the approach scale to the size of the cluster?

Research Question 3 – Does the approach scale with the size of the input?






Evaluation - Reproducibility

Research Question 4 – Using their infrastructure, can researchers easily reproduce previously published results?


Evaluation - Reproducibility

Conducted controlled experiment Selected group of 8 researchers Each chose 3 tasks


References

http://design.cs.iastate.edu/papers/ICSE-13/icse13.pdf

http://boa.cs.iastate.edu/docs/




http://boa.cs.iastate.edu/docs/


Documents

Presenter: Joshan V John Robert Dyer, Hoan Anh Nguyen, Hridesh Rajan & Tien N. Nguyen Iowa State University, USA Instructor: Christoph Csallner 1 Joshan