SLIDE 1
Lecture 3 Guiding Software Development Where do you go next? - - PowerPoint PPT Presentation
Lecture 3 Guiding Software Development Where do you go next? - - PowerPoint PPT Presentation
Lecture 3 Guiding Software Development Where do you go next? Mentoring! Evolutionay Coupling Research Hypothesis Evolutionay Coupling eROSE Reengineering of Software Evolution Tom Zimmermann Saarland University Peter Weigerber
SLIDE 2
SLIDE 3
Mentoring!
SLIDE 4
Research Hypothesis Evolutionay Coupling
SLIDE 5
Evolutionay Coupling
SLIDE 6
eROSE
Tom Zimmermann • Saarland University Peter Weißgerber • University of Trier Stephan Diehl • University of Trier Andreas Zeller • Saarland University
Reengineering of Software Evolution
SLIDE 7
SLIDE 8
Developers who changed this function also changed...
SLIDE 9
eROSE: Guiding Developers
Purchase History
Customers who bought this item also bought...
Version Archive
Developers who changed this function also changed...
SLIDE 10
What Can eROSE do?
Suggest and predict likely changes
SLIDE 11
What Can eROSE do?
Prevent errors due to incomplete changes.
SLIDE 12
What Can eROSE do?
Detect coupling undetectable by program analysis.
SLIDE 13
Evolutionary Coupling in GNU DDD
SLIDE 14
ddd directory Icons Libraries Patches Tests
Evolutionary Coupling in GNU DDD
SLIDE 15
Evolutionary Coupling
SLIDE 16
What can eROSE do?
SLIDE 17
SLIDE 18
SLIDE 19
User inserts a new preference into the field fkeys[].
SLIDE 20
User inserts a new preference into the field fkeys[].
SLIDE 21
User inserts a new preference into the field fkeys[]. User inserts a new preference into the field fkeys[].
SLIDE 22
SLIDE 23
eROSE suggests further locations.
SLIDE 24
http://www.st.cs.uni-sb.de/softevo/erose/
SLIDE 25
SLIDE 26
eROSE prevents incomplete changes.
SLIDE 27
eROSE Architecture
- 1. Preprocess CVS Archive.
- 2. Mine association rules between entities.
SLIDE 28
Processing CVS data
SLIDE 29
Processing CVS data
SLIDE 30
Processing CVS data
- 1. Comparing files
- 2. Building transactions
SLIDE 31
Mapping Changes to Entities
r1 r2
SLIDE 32
Mapping Changes to Entities
A() C() E() D() B()
r1 r2
Parse r1 for entities
SLIDE 33
Mapping Changes to Entities
A() C() E() D() B() A() B() E() F() D()
r1 r2
Parse r1 for entities Parse r2 for entities
SLIDE 34
Mapping Changes to Entities
A() C() E() D() B() A() B() E() F() D()
r1 r2
Parse r1 for entities Parse r2 for entities diff diff diff diff gone new
SLIDE 35
CVS Transactions
New Developer
SLIDE 36
CVS Transactions
New Developer
user changes files
SLIDE 37
CVS Transactions
Project Files (changed)
New Developer
user changes files local copy changed
SLIDE 38
CVS Transactions
Project Files (changed)
New Developer
user changes files local copy changed
CVS Server
Project Files (changed)
local copy changed
SLIDE 39
CVS Transactions
Project Files (changed)
New Developer
user changes files local copy changed
CVS Server
Project Files (changed)
local copy changed
T r a n s a c t i
- n
SLIDE 40
Infer Transactions: Time Windows
All changes by the same developer, with the same message, made at the same time belong to one transaction.
A:1.3 B:1.2 C:1.4 D:1.3 E:1.5 same author + same log message
Fixed Time Window ∀δi : ∀δj : | time (δi) − time (δj) | ≤ T
A:1.3 B:1.2 C:1.4 D:1.3 E:1.5 same author + same log message
Sliding Time Window ∀δi : ∃δj : | time (δi) − time (δj) | ≤ T
200 seconds 200 seconds
SLIDE 41
Infer Transactions: Time Windows
All changes by the same developer, with the same message, made at the same time belong to one transaction.
A:1.3 B:1.2 C:1.4 D:1.3 E:1.5 same author + same log message
Fixed Time Window ∀δi : ∀δj : | time (δi) − time (δj) | ≤ T
A:1.3 B:1.2 C:1.4 D:1.3 E:1.5 same author + same log message
Sliding Time Window ∀δi : ∃δj : | time (δi) − time (δj) | ≤ T
200 seconds 200 seconds
SLIDE 42
Building Transactions
CVS
150,000
SLIDE 43
Building Transactions
CVS
150,000
createGeneralPage() createT extComparePage() fKeys[] initDefaults() buildnotes_compare.html PatchMessages.properties plugin.properties
2003-02-19 (aweinand): fixed #13332
SLIDE 44
Building Transactions
CVS
150,000
createGeneralPage() createT extComparePage() fKeys[] initDefaults() buildnotes_compare.html PatchMessages.properties plugin.properties
2003-02-19 (aweinand): fixed #13332
same author + message + time
SLIDE 45
From Transactions to Rules
{alter(field, fkeys[],...)} ⇒ {
}
alter(method, initDefaults(),...), alter(file, plug.properties,...),
Example Rule association rule r is a pair (x1, x2) where x1 is the antecedent and x2 is the consequent
SLIDE 46
Entities
An entity is a triple (c, i, p) where c is the syntactic category, i is the identifier p is the parent entity
SLIDE 47
Entities
An entity is a triple (c, i, p) where c is the syntactic category, i is the identifier p is the parent entity Example (method, initDefaults(), (class, Comp, (file, Comp.java, ...))) c i p
SLIDE 48
Types of Operations
SLIDE 49
Types of Operations
alter(e)
SLIDE 50
Types of Operations
alter(e) add_to(e)
SLIDE 51
Types of Operations
alter(e) add_to(e) del_from(e)
SLIDE 52
Mining Associations
SLIDE 53
Mining Associations
EROSE finds past transactions
SLIDE 54
Mining Associations
fKeys[] initDefaults() ... plugin.properties
#104223
fKeys[] initDefaults() ... plugin.properties
#756
fKeys[] initDefaults() ... plugin.properties
#6721
fKeys[] initDefaults() ... plugin.properties
#21078
fKeys[] initDefaults() ... plugin.properties
#42432
fKeys[] initDefaults() ... plugin.properties
#51345
fKeys[] initDefaults() ... plugin.properties
#59998
fKeys[] initDefaults() ... plugin.properties
#71003
fKeys[] initDefaults() ...
#87264
fKeys[] initDefaults() ... plugin.properties
#91220
fKeys[] initDefaults() ... plugin.properties
#101823
EROSE finds past transactions
SLIDE 55
SLIDE 56
SLIDE 57
Support & Confidence
SLIDE 58
Support & Confidence
SLIDE 59
Support & Confidence
transactions where fkeys[] was altered
SLIDE 60
Support & Confidence
transactions where fkeys[] was altered transactions where initDefaults() and plug.properties were altered
SLIDE 61
Support & Confidence
Support = 11 Confidence = 10/11 = 0.909
transactions where fkeys[] was altered transactions where initDefaults() and plug.properties were altered
SLIDE 62
Applying Rules
Situation ∑ = {alter(field, fkeys[],...)} apply (∑)
R (∑⇒x2) ∈ R
= ∪ x2
i.e., the set of suggestions for a situation ∑ and rules R is defined as the union of the consequents of all matching rules.
{
}
alter(method, initDefaults(),...), alter(file, plug.properties,...),
apply (∑) =
R
SLIDE 63
Applying Rules
{
}
alter(field, keys[],...), alter(method, init.Defaults(),...)
∑ =
{
}
alter(file, plug.properties(),...), alter(file, build.html,...),
apply (∑) =
R
SLIDE 64
Applying Rules
{
}
alter(field, keys[],...), alter(method, init.Defaults(),...)
∑ =
{
}
alter(file, plug.properties(),...), alter(file, build.html,...),
apply (∑) =
R
eRose can build rules on the fly.
SLIDE 65
Computing Association Rules
eROSE uses the Apriori Algorithm to compute rules. Takes several days to generate rule set, but eROSE has
- ptimisations built it.
- 1. Constrained Antecedents (builds rules on the fly)
- 2. Single Consequents (∑⇒{e})
SLIDE 66
Some Rule Examples
Coupling in GCC Compiler {alter(type, processor_cost (file, i386.h,...))} ⇒ {alter(var, i386_cost, (file, i386.c,...))} {alter(var, i486_cost, (file, i386.c,...))} {alter(var, k6_cost, (file, i386.c,...))} {alter(var, pentium_cost, (file, i386.c,...))} {alter(var, pentiumpro_cost, (file, i386.c,...))} ∑ =
SLIDE 67
Some Rule Examples
PostGRES Documentation {alter(file, createuser.sgml, ...)), ⇒ ∑ = {alter(file, dropuser.sgml, ...)), {alter(file, createdb.sgml, ...)), {alter(file, dropdb.sgml, ...)),
SLIDE 68
PostgreSQL
Evaluation
jEdit KOffice GIMP
SLIDE 69
PostgreSQL
Evaluation
jEdit KOffice GIMP EROSE predicts 33% of all changed entities. (files: 44%)
SLIDE 70
PostgreSQL
Evaluation
jEdit KOffice GIMP EROSE predicts 33% of all changed entities. (files: 44%) In 70% of all transactions, EROSE’s topmost three suggestions contain a changed entity.
SLIDE 71
PostgreSQL
Evaluation
jEdit KOffice GIMP EROSE predicts 33% of all changed entities. (files: 44%) In 70% of all transactions, EROSE’s topmost three suggestions contain a changed entity. EROSE learns quickly (within 30 days).
SLIDE 72
History Required
SLIDE 73
Software Development Artefacts
SLIDE 74
Software Development Artefacts
Source Code
SLIDE 75
Software Development Artefacts
Email Source Code
SLIDE 76
Software Development Artefacts
Bug Database Email Requirements CVS Archive Discussion Forums Test Suites Source Code
SLIDE 77
Software Development Artefacts
Bug Database Email Requirements CVS Archive Discussion Forums Test Suites Source Code
SLIDE 78
Artefact Types
Source File Versions Change Tasks Messages Other Documents
SLIDE 79
Relationship Types
File Versions Change Tasks
Messages
Documents
similar to documents check-in package reply to similar to writes writes posts works on implements about
SLIDE 80
Relationship Types
File Versions Change Tasks
Messages
Documents
similar to documents check-in package reply to similar to writes writes posts works on implements about
SLIDE 81
Relationship Types
File Versions Change Tasks
Messages
Documents
similar to documents check-in package reply to similar to writes writes posts works on implements about
SLIDE 82
Relationship Types
File Versions Change Tasks
Messages
Documents
similar to documents check-in package reply to similar to writes writes posts works on implements about
Hipikat infers links by combining information contained within project artefacts and metainformation about artefacts from different information sources.
SLIDE 83
Relationship Types
File Versions Change Tasks
Messages
Documents
similar to documents check-in package reply to similar to writes writes posts works on implements about
Hipikat infers links by combining information contained within project artefacts and metainformation about artefacts from different information sources.
SLIDE 84
Hipikat Client-Server Protocol
SLIDE 85
Hipikat Client-Server Protocol
Artefact Database
SLIDE 86
Hipikat Client-Server Protocol
Artefact Database
web
Mail/News
CVS
SLIDE 87
Hipikat Client-Server Protocol
Artefact Database
web
Mail/News
CVS
Update
SLIDE 88
Hipikat Client-Server Protocol
Artefact Database
web
Mail/News
CVS
Update
monitors
SLIDE 89
Hipikat Client-Server Protocol
Artefact Database
web
Mail/News
CVS
Update Insert artefact
monitors
SLIDE 90
Hipikat Client-Server Protocol
Artefact Database
web
Mail/News
CVS
Identification Update Insert artefact
monitors
SLIDE 91
Hipikat Client-Server Protocol
Artefact Database
web
Mail/News
CVS
Identification Listen for new artefacts Update Insert artefact
monitors
SLIDE 92
Hipikat Client-Server Protocol
Artefact Database
web
Mail/News
CVS
Identification Listen for new artefacts Update Inserts links Insert artefact
monitors
SLIDE 93
Hipikat Client-Server Protocol
Artefact Database
web
Mail/News
CVS
Identification Listen for new artefacts Update Inserts links Insert artefact
monitors
SLIDE 94
Hipikat Client-Server Protocol
Artefact Database
web
Mail/News
CVS
Identification Listen for new artefacts Update Inserts links Insert artefact
Query monitors
SLIDE 95
Hipikat Client-Server Protocol
Artefact Database
web
Mail/News
CVS
Identification Listen for new artefacts Selection Update Inserts links Insert artefact
Query monitors
SLIDE 96
Hipikat Client-Server Protocol
Artefact Database
web
Mail/News
CVS
Identification Listen for new artefacts Selection Follows links Update Inserts links Insert artefact
Query monitors
SLIDE 97
Hipikat Client-Server Protocol
Artefact Database
web
Mail/News
CVS
Identification Listen for new artefacts Selection Follows links Update Inserts links Insert artefact
Query
Related artefacts
monitors
SLIDE 98
Update
Update
- has a separate sub-modules to handle different
types of data, e.g. CVS and Bugzilla.
- New and changed artefacts are inserted into
Artefact database.
- Change listeners are notified.
SLIDE 99
Identification
Identification
SLIDE 100
Identification
Identification Log Matcher
SLIDE 101
Identification
Identification Activity Matcher Log Matcher
SLIDE 102
Identification
Identification Activity Matcher Text Similarity Matcher Log Matcher
SLIDE 103
Identification
Identification Activity Matcher Text Similarity Matcher Log Matcher CVS Check-in Package Matcher
SLIDE 104
Identification
Identification Activity Matcher Text Similarity Matcher Log Matcher CVS Check-in Package Matcher
Thread Matcher
SLIDE 105
Selection
Selection
SLIDE 106
Hipikat Client (Eclipse Plug-in)
SLIDE 107
Evaluation
The Eclipse Newcomer Study
Newcomers
- Is the tool helpful to them?
- When and from which artefacts will they query?
- How to they evaluate and utilise recommendations?
SLIDE 108
Evaluation
The Eclipse Newcomer Study
Newcomers
- Is the tool helpful to them?
- When and from which artefacts will they query?
- How to they evaluate and utilise recommendations?
Experienced
SLIDE 109
Summary
eROSE: Guiding Developers
Purchase History
Customers who bought this item also bought...
Version Archive
Developers who changed this function also changed...
Hipikat Client-Server Protocol
Artefact Database
web Mail/NewsCVS
Identification Listen for new artefacts Selection Follows links Update Inserts links Insert artefact Query Related artefacts monitorsA Recommender System A Recommender System?