With mining software repositories (MSR), we analyze the rich data created during the whole evolution of one or more software projects. One major obstacle in MSR is the heterogeneity and complexity of source code as a data source. With model-based technology in general and reverse engineering in particular, we can use abstraction to overcome this obstacle. But, this raises a new question: can we apply existing reverse engineering frameworks that were designed to create models from a single revision of a software system to analyze all revisions of such a system at once? This paper presents a framework that uses a combination of EMF, the reverse engineering framework Modisco, a NoSQL-based model persistence framework, and OCL-like expressions to create and analyze fully resolved AST-level model representations of whole source code repositories. We evaluated the feasibility of this approach with a series of experiments on the Eclipse code-base.

1. Model-based Analysis of
Source Code Repositories
Markus Scheidgen
1
Martin Schmidt Joachim Fischer
{scheidge,schmidma,fischer@informatik.hu-berlin.de
Humboldt Universität zu Berlin

2. Agenda
▶ Software Evolution, Reverse Engineering, and Mining Software
Repositories
▶ Model-based Mining of Software Repositories
▶ srcrepo a framework for model-bases analysis of software
repositories
▶ Experiments with Eclipse’s software repositories
▶ Conclusions
2

3. software maintenance
RHEAD
…
R0
■ quality assessment
■ implicit dependencies
software engineering
Software Evolution
3
requirements
M
M {C}
S
user
problem
M
M {C}
S
user
mining software repositories1
software modernization
OMG’s ADM
(Achitecture-Driven Modernization)
‣AST Meta-Model (ASTM)
‣Knowledge Discovery Meta-
Model (KDM)
‣Software Metrics Meta-Model
(SMM)
M
SS
M
M
M
{C} {C}
reverseengineering
forwardengineering
transformation
1.H. Kagdi, M.L. Collard, J.I. Maletic: A survey and taxonomy of approaches for mining software repositories in the context of software
evolution; Journal of Software Maintenance and Evolution: Research and Practice; Vol.19/Nr.2/2007

4. Mining Software Repositories – In General
▶ The term mining software repositories (MSR) has been coined
to describe a broad class of investigations into the examination
of software repositories.
▶ The premise of MSR is that empirical and systematic
investigations of repositories will shed new light on the process
of software evolution. [1]
▶ Different scopes, e.g. single software projects vs. many
software projects
▶ Different goals, e.g. quality assessments and implicit
dependencies vs. generalizations about software evolution
4
1.H. Kagdi, M.L. Collard, J.I. Maletic: A survey and taxonomy of approaches for mining software repositories in the context of software
evolution; Journal of Software Maintenance and Evolution: Research and Practice; Vol.19/Nr.2/2007

5. Model-based Mining of Software Repositories
5
MS M{C} reverse engineering
1.E.J. Chikofsky, J.H. Cross: Reverse engineering and design recovery: A taxonomy; IEEE Software; Vol.7/Nr.1/1990
2.R. Lincke, J. Lundberg, W. Löwe: Comparing Software Metrics Tools; 8th International Symp. on Software Testing and Analysis; 2008

6. Model-based Mining of Software Repositories
5
MS M{C}
MM{Cn}
RHEAD
…
R
0
reverse engineering
1.E.J. Chikofsky, J.H. Cross: Reverse engineering and design recovery: A taxonomy; IEEE Software; Vol.7/Nr.1/1990
2.R. Lincke, J. Lundberg, W. Löwe: Comparing Software Metrics Tools; 8th International Symp. on Software Testing and Analysis; 2008

7. Model-based Mining of Software Repositories
5
MS M{C}
MM{Cn}
RHEAD
…
R
0
{Cn-1} MM
reverse engineering
1.E.J. Chikofsky, J.H. Cross: Reverse engineering and design recovery: A taxonomy; IEEE Software; Vol.7/Nr.1/1990
2.R. Lincke, J. Lundberg, W. Löwe: Comparing Software Metrics Tools; 8th International Symp. on Software Testing and Analysis; 2008

8. Model-based Mining of Software Repositories
5
MS M{C}
MM{Cn}
RHEAD
…
R
0
{Cn-1} MM
{C0}
…
MM
…
…
…
reverse engineering
1.E.J. Chikofsky, J.H. Cross: Reverse engineering and design recovery: A taxonomy; IEEE Software; Vol.7/Nr.1/1990
2.R. Lincke, J. Lundberg, W. Löwe: Comparing Software Metrics Tools; 8th International Symp. on Software Testing and Analysis; 2008

9. Model-based Mining of Software Repositories
5
MS M{C}
MM{Cn}
RHEAD
…
R
0
{Cn-1} MM
{C0}
…
MM
…
…
…
reverse engineering
1.E.J. Chikofsky, J.H. Cross: Reverse engineering and design recovery: A taxonomy; IEEE Software; Vol.7/Nr.1/1990
2.R. Lincke, J. Lundberg, W. Löwe: Comparing Software Metrics Tools; 8th International Symp. on Software Testing and Analysis; 2008

10. Model-based Mining of Software Repositories
▶ Scope
■ depends on concreter MSR-application and its goals
■ number of software projects: single repositories, large
repositories, ultra-large repositories
■ Sources as text and text based metrics, e.g. LOC
■ Declarations only: packages, classes, methods, but no
statements, expressions, etc.
■ Full AST with or without cross-references
6

11. Model-based Mining of Software Repositories
▶ Scope
■ depends on concreter MSR-application and its goals
■ number of software projects: single repositories, large
repositories, ultra-large repositories
■ Sources as text and text based metrics, e.g. LOC
■ Declarations only: packages, classes, methods, but no
statements, expressions, etc.
■ Full AST with or without cross-references
7

12. Model-based Mining of Software Repositories
▶ MSR tools are already “model-based”, but in a proprietary
manner
▶ Idea: existing reverse engineering framework and
corresponding standard meta-models and modeling
frameworks instead of proprietary solutions
▶ Goals
■ deal with heterogeneity (different version control systems,
different languages)
■ reuse of existing meta-models, transformations, and languages
■ interoperability with existing analysis tools
■ retaining meaningful scalability
8

13. Model-based MSR Strategies
9
versioncontrolsystem
A1-A2
A1 B1
B1-B3

14. Model-based MSR Strategies
9
snapshot
A1 B1
Checkout(r)
versioncontrolsystem
A1-A2
A1 B1
B1-B3

15. Model-based MSR Strategies
9
snapshot
A1 B1
Checkout(r)
versioncontrolsystem
A1-A2
A1 B1
B1-B3
snapshot
A2 B1

16. snapshot
A2 B3
Model-based MSR Strategies
9
snapshot
A1 B1
Checkout(r)
versioncontrolsystem
A1-A2
A1 B1
B1-B3
snapshot
A2 B1

17. snapshot
A2 B3
Model-based MSR Strategies
9
snapshot
A1 B1
Checkout(r)
versioncontrolsystem
A1-A2
A1 B1
B1-B3
snapshot
A2 B1
snapshot
X
d2CUs(r)
Parse(d)

18. snapshot
A2 B3
Model-based MSR Strategies
9
snapshot
A1 B1
Checkout(r)
versioncontrolsystem
A1-A2
A1 B1
B1-B3
snapshot
A2 B1
snapshot
X
d2CUs(r)
Parse(d)
snapshot

19. snapshot
snapshot
A2 B3
Model-based MSR Strategies
9
snapshot
A1 B1
Checkout(r)
versioncontrolsystem
A1-A2
A1 B1
B1-B3
snapshot
A2 B1
snapshot
X
d2CUs(r)
Parse(d)
snapshot

20. snapshot
snapshot
A2 B3
Model-based MSR Strategies
9
snapshot
A1 B1
Checkout(r)
versioncontrolsystem
A1-A2
A1 B1
B1-B3
snapshot
A2 B1
snapshot
X
d2CUs(r)
Parse(d)
snapshot
M1
Analysis(r)

21. snapshot
snapshot
A2 B3
Model-based MSR Strategies
9
snapshot
A1 B1
Checkout(r)
versioncontrolsystem
A1-A2
A1 B1
B1-B3
snapshot
A2 B1
snapshot
X
d2CUs(r)
Parse(d)
snapshot
M1
Analysis(r)
M2

22. snapshot
snapshot
A2 B3
Model-based MSR Strategies
9
snapshot
A1 B1
Checkout(r)
versioncontrolsystem
A1-A2
A1 B1
B1-B3
snapshot
A2 B1
snapshot
X
d2CUs(r)
Parse(d)
snapshot
M1
Analysis(r)
M3
M2

23. snapshot
snapshot
A2 B3
Model-based MSR Strategies
9
snapshot
A1 B1
Checkout(r)
versioncontrolsystem
A1-A2
A1 B1
B1-B3
snapshot
A2 B1
snapshot
X
d2CUs(r)
Parse(d)
snapshot
M1
Analysis(r)
M3
M2
X
R
Checkout +
X
CUs
Parse + Analysis
!
X
R
Checkout +
X
CUs
(Parse + Merge) + Analysis
!
X
R
X
CUs
(Load + Merge) + Analysis
!
X X

24. snapshot
snapshot
A2 B3
Model-based MSR Strategies
10
snapshot
A1 B1
Checkout(r)
versioncontrolsystem
A1-A2
A1 B1
B1-B3
snapshot
A2 B1
snapshot
X
d2CUs(r)
Parse(d)
snapshot
M1
Analysis(r)
M3
M2

25. snapshot
snapshot
A2 B3
Model-based MSR Strategies
10
snapshot
A1 B1
Checkout(r)
versioncontrolsystem
A1-A2
A1 B1
B1-B3
snapshot
A2 B1
snapshot
X
d2CUs(r)
Parse(d)
snapshot
M1
Analysis(r)
M3
M2
f
B.f
fB.f
Parse(d)
X
d2 CUs(r)

26. snapshot
snapshot
A2 B3
Model-based MSR Strategies
10
snapshot
A1 B1
Checkout(r)
versioncontrolsystem
A1-A2
A1 B1
B1-B3
snapshot
A2 B1
snapshot
X
d2CUs(r)
Parse(d)
snapshot
M1
Analysis(r)
M3
M2
Merge(r)
f
B.f
fB.f
Parse(d)
X
d2 CUs(r)

27. snapshot
snapshot
A2 B3
Model-based MSR Strategies
10
snapshot
A1 B1
Checkout(r)
versioncontrolsystem
A1-A2
A1 B1
B1-B3
snapshot
A2 B1
snapshot
X
d2CUs(r)
Parse(d)
snapshot
M1
Analysis(r)
M3
M2
Merge(r)
f
B.f
fB.f
Parse(d)
X
d2 CUs(r)
X
R
Checkout +
X
CUs
Parse + Analysis
!
X
R
Checkout +
X
CUs
(Parse + Merge) + Analysis
!
X X
!

28. snapshot
snapshot
A2 B3
Model-based MSR Strategies
11
snapshot
A1 B1
Checkout(r)
versioncontrolsystem
A1-A2
A1 B1
B1-B3
snapshot
A2 B1
snapshot
X
d2CUs(r)
Parse(d)
snapshot
M1
Analysis(r)
M3
M2
Merge(r)
f
B.f
fB.f
Parse(d)
X
d2 CUs(r)

29. snapshot
snapshot
A2 B3
Model-based MSR Strategies
11
snapshot
A1 B1
Checkout(r)
versioncontrolsystem
A1-A2
A1 B1
B1-B3
snapshot
A2 B1
snapshot
X
d2CUs(r)
Parse(d)
snapshot
M1
Analysis(r)
M3
M2
Merge(r)
f
B.f
fB.f
Parse(d)
X
d2 CUs(r)
Load(r)Save(r)

30. snapshot
snapshot
A2 B3
Model-based MSR Strategies
11
snapshot
A1 B1
Checkout(r)
versioncontrolsystem
A1-A2
A1 B1
B1-B3
snapshot
A2 B1
snapshot
X
d2CUs(r)
Parse(d)
snapshot
M1
Analysis(r)
M3
M2
Merge(r)
f
B.f
fB.f
Parse(d)
X
d2 CUs(r)
Load(r)Save(r)
X
R
Checkout +
X
CUs
Parse + Analysis
!
X
R
Checkout +
X
CUs
(Parse + Merge) + Analysis
!
X
R
X
CUs
(Load + Merge) + Analysis
!
X
R
X
CUs
(Load + Analysis0
)

31. snapshot
snapshot
A2 B3
Model-based MSR Strategies
12
snapshot
A1 B1
Checkout(r)
versioncontrolsystem
A1-A2
A1 B1
B1-B3
snapshot
A2 B1
snapshot
X
d2CUs(r)
Parse(d)
snapshot
M1
Analysis(r)
M3
M2
Merge(r)
f
B.f
fB.f
Parse(d)
X
d2 CUs(r)
Load(r)Save(r)
X
R
Checkout +
X
CUs
Parse + Analysis
!
X
R
Checkout +
X
CUs
(Parse + Merge) + Analysis
!
X
R
X
CUs
(Load + Merge) + Analysis
!
X
R
X
CUs
(Load + Analysis0
)
Analysis0
(r)

32. snapshot
snapshot
A2 B3
Model-based MSR Strategies
12
snapshot
A1 B1
Checkout(r)
versioncontrolsystem
A1-A2
A1 B1
B1-B3
snapshot
A2 B1
snapshot
X
d2CUs(r)
Parse(d)
snapshot
M1
Analysis(r)
M3
M2
Merge(r)
f
B.f
fB.f
Parse(d)
X
d2 CUs(r)
Load(r)Save(r)
Analysis0
(r)
X
R
Checkout +
X
CUs
Parse + Analysis
!
X
R
Checkout +
X
CUs
(Parse + Merge) + Analysis
!
X
R
X
CUs
(Load + Merge) + Analysis
!
X
R
X
CUs
(Load + Analysis0
)

33. snapshot
snapshot
A2 B3
Model-based MSR Strategies
13
snapshot
A1 B1
Checkout(r)
versioncontrolsystem
A1-A2
A1 B1
B1-B3
snapshot
A2 B1
snapshot
X
d2CUs(r)
Parse(d)
snapshot
M1
Analysis(r)
M3
M2
Merge(r)
f
B.f
fB.f
Parse(d)
X
d2 CUs(r)
Load(r)Save(r)
Analysis0
(r)

34. Research Questions
▶ Assumptions
■ Development of MSR-applications based on models,
transformation languages and standardized meta-models is
favorable
■ Some MSR-applications need to analyze source code on a deep
(AST) level
■ MSR-analysis is performed iteratively
▶ Hypotheses
■ Models of source code repositories can be created and persisted
■ Traversing existing persistent models of source code repositories
is much faster than traversing transient models that are created
from version control system on the fly
14

35. srcrepo – A Framework for Model-based MSR
▶ Eclipse’s MoDisco as reverse engineering framework
■ reverse engineering for Java, based on EMF
■ Support for many JRE-ased languages: Java, xText, JSP, XML
■ creates instances of a Java EMF meta-model that corresponds to the handwritten
JDT AST-model
■ provides transformation to language independent artifacts, e.g. KDM
▶ EMF-Fragments1 to store very large-models
■ uses No-SQL databases and stores larger model fragments within database entries
■ in contrast to object-by-object stores such as ORM-based CDO or No-SQL-based
Morsa or Neo4J
▶ Xtend programming with higher order functions to mimic OCL-style
definition of software metrics2
15
1.M.Scheidgen, A.Zubow,J.Fischer,T.H.Kolbe: Automated and Transparent Model Fragmentation for Persisting Large Models; ACM/IEEE
15th International Conference on Model Driven Engineering Languages & Systems (MODELS); Innsbruck; 2012
2. M.Scheidgen, J.Fischer: Model-based Mining of Software Repositories; 8th Systems and Modeling Conference, Valencia, Spain,
September 29th, 2014

36. “OCL” to Calculate Metrics of AST-Models
16
// Weighted number of methods per class.
def wmc(AbstractTypeDeclaration type,(Block) int weight) {
type.bodyDeclarations.sum[if (body != null) weight.apply(it.body) else 1]
}

37. Experiments
▶ Eclipse Foundation sources, i.e. Eclipse platform and plug-
ins (large scale software repository)
▶ Organized in different (couple hundred) projects: jdt, cdt,
emf, ...
▶ Available via GITHub
▶ GIT repositories can be gathered automated via GITHub’s
REST-ful API
▶ 200 largest Eclipse repositories that actually contained Java
code: 6.6 GB Git, 400 MLOC, 250 GB model with 4 billion
objects.
17

38. Example Plot: Halstead-length for each Revision; Eclipse CDT
18
2004 2006 2008 2010 2012 2014 2016
0246810
time (years)
WMCwithHalsteadlength(x106
)
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+++++++++++++++++++++++++++++++++++++++
++++++++++++++++++++++++++++++++++++++++++++++++++++++
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ ++++++++++
+++++
+++++ ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+++++
++++++++
+++++++++++++++++++++
++++++++++++++++++++++++++++++++++++
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+
++++
++++
++++++++++++++++++ ++
++++++++++++++++++
++++++++++++++++++++ ++++ +
+++++++++++++ ++
+++++++++++++++++
+
++++++++ +
++++++++
++++++++
++++++++++ +
+
++++
++
+++
+ ++ + +++++ +++++++++++++++++++++++++++++++++++++++

39. Model Create v Analysis Times
19
jdt.ui
xtext
eclipselink
jdt.core
swt
cdt
ocl
ptp
org.aspectj
cdo
udf
merge/increment
load
save
parse
checkout
time(hours)
0
2
4
6
8
10
12
14
checkout parse save load merge/
increment
udf
0200400600800
avgtimeperrevision(ms)

40. jdt.core
cdt
jdt.ui
cdo
f.emfstore.core
emf
jdt.debug
emf.compare
emf.texo
.diffmerge.core
GIT size
Model size
GIT repository vs model size
inGB
0
5
10
15
Diskspace
20

41. Delta-Compression1
21
cdt
webtools
gmf−tooling
rap
jdt.core
delta-models
initial revisions
uncompressed
Named element matching
GB
0
2
4
6
8
10
12
14
cdt
webtools
gmf−tooling
rap
jdt.core
detla-models
initial revisions
uncompressed
Meta-class matching
GB
0
2
4
6
8
10
12
14
cdt
webtools
gmf−tooling
rap
jdt.core
delta-lines
initial revisions
uncompressed
Line matching
MLines
0
20
40
60
80
100
initial
revisions
delta
models for
meta-class
delta
lines
delta models
for named
elements
020406080
Compressed relative to full size
(%)
parse compress
named
elements
compress
meta-class
decompr.
meta-class
decompr.
named
elements
parse compress
named
elements
compress
meta-class
decompr.
meta-class
decompr.
named
elements
050010001500
Avg. execution times
avg.timeperrevision(ms)
10502001000
Avg. execution times (logarithmic)
avg.timeperrevision(ms)
1.M.Scheidgen: Evaluation of Model Comparison for Delta-Compression in Model Persistence; BigMDE 2016 (at STAF 2016),Vienna,
Austria, July 6-7, 2016

42. Model Creation v Analysis with Delta-Compression
22
jdt.ui
xtext
eclipselink
jdt.core
swt
cdt
ocl
ptp
org.aspectj
cdo
udf
merge/increment
load
save
compress
parse
checkout
time(hours)
0
2
4
6
8
10
12
14
jdt.ui
xtext
eclipselink
jdt.core
swt
cdt
ocl
ptp
org.aspectj
cdo
udf
merge/increment
load
save
parse
checkout
time(hours)
0
2
4
6
8
10
12
14
without compression with compression

43. Conclusions
▶ MSR can support software evolution and helps to
understand software evolution
▶ Traversing a source code repository to gather information
(MSR) is very time consuming, especially with iterative
analysis
▶ It is possible to save most of this time via saving data in its
model state, at the cost of comparably large models that
need to be persisted
▶ The MSR analysis execution time savings are considerable
23