Presented at MSR 2019 https://doi.org/10.1109/MSR.2019.00083 https://arxiv.org/abs/1904.01221

1. TheImpactof
Systematic Edits in
History Slicing
Tokyo Institute ofTechnology, Japan
Ryosuke
Funaki
Shinpei
Hayashi
Motoshi
Saeki
[ Chronology by Les Chatfield: http://www.flickr.com/photos/elsie/4607687530/ ]

2. TheImpactof Systematic Edits
inHistory Slicing
2
Similar edits applied to multiple
locations of source code
Extracting sub-history based on
dependency among change elements

3. Motivation:
Sub-history Extraction
3
Branch Refactoring Backporting
4
3
2
1
4
ver2
2
4
3
2
1
4
2
3
1
PR1 PR2develop master
Ver. 2 release
Related
change
Security
fix

4. History
Slicing
4
[ Chain by rob patrick: https://www.flickr.com/photos/alkalinezoo/1473626338/
Line-based
[Servant et al.]
Edit-based [Maruyama et al.]
Commit-based [Li et al.]

5. 5
developers often perform
systematic edits—similar,
but not identical, changes to
many locations.
’’
Meng et al.,“LASE: Locating and Applying Systematic Edits by Learning from Examples” Proc. ICSE 2013.
[ 3/4" 285HV Grid 1749 by Peter Kemmer: https://www.flickr.com/photos/pkmousie/2225340795/ ]

6. 6https://github.com/apache/commons-collections/commit/059c468
- long offset
+ final long offset

7. Systematic edits lead to enlarging
commit-based history slices
7
Slicing criterion
Systematic edits
Changes in files are
semantically related
Changes in files are
independent
Irrelevant Irrelevant
- long offset
+ final long offset

8. How much do
systematic-edit-
based commits
impact
the size of
commit-based
history slices?
8
Research
Question
[ protractors by Dean Hochman: https://www.flickr.com/photos/deanhochman/14481958642/ ]

9. 9
Dependency among Changed Files
l Introduces “Commit dependency”
l Remove them if the commit is regarded as systematic
– Systematic edits are detected using an extended version of
technique by Molderez et al.
Slicing criterion
Systematic edits
Molderez et al.:“Mining change histories for unknown systematic edits,” Proc. MSR 2017.

10. Detecting Systematic Edits
Arranged the technique by Molderez et al.
• Calculates AST edit scripts
• Detects systematic edits if the script pattern for each
method is all the same
10Molderez et al.:“Mining change histories for unknown systematic edits,” Proc. MSR 2017.
BA BA
INS final
INS final
INS final
INS final
: Method
A B

11. Empirical Study: Setup
11
4.0 – 4.1 (164 slices)
4.1 – 4.2 ( 83 slices)
3.3 – 3.4 (211 slices)
9
8
6
7
8
6
9
77
6
9つの部分履歴
9コミット
Extracted slices using each commit in the
original history as a slicing criterion
Original
history
Slicing
Slicing
Slicing

12. RQ: Slice size when splitting commits
consisting of systematic edits
12
Collections 4.0-4.1 Collections 4.1-4.2 Net 3.3-3.4
0
10
20
30
40
50
60
70
80
90
0 10 20 30 40 50 60 70 80
Origin al
Redu ced
Size
0
5
10
15
20
25
30
35
40
0 10 20 30
Origin al
Redu ced
Slices
0
20
40
60
80
100
120
140
0 10 20 30 40 50 60
Size
The size of history slices decreased by
13.3-57.2% on average when splitting
commits consisting of systematic edits.

13. TheImpactof
Systematic Edits in
History Slicing
RQ: Slice size when splitting commits
consisting of systematic edits
15
Collections 4.0-4.1 Collections 4.1-4.2 Net 3.3-3.4
0
10
20
30
40
50
60
70
80
90
0 10 20 30 40 50 60 70 80
Origin al
Redu ced
Size
0
5
10
15
20
25
30
35
40
0 10 20 30
Origin al
Redu ced
Slices
0
20
40
60
80
100
120
140
0 10 20 30 40 50 60 70
Size
The size of history slices decreased by
13.3-57.2% on average when splitting
commits consisting of systematic edits.10
Dependency among Changed Files
l Introduces “Commit dependency”
l Remove them if the commit is regarded as systematic
Slicing criterion
Systematic edits
[ Chronology by Les Chatfield: http://www.flickr.com/photos/elsie/4607687530/ ]