Lecture 6 & 7. Empirical Studies of Software Evolution: Code Decay. EE382V Software Evolution: Spring 2009, Instructor Miryung Kim

Transcription

1 Lecture 6 & 7 Empirical Studies of Software Evolution: Code Decay

2 Announcement Your proposals have been graded. Literature Survey & Tool Evaluation: Each in the range of 1-4 Project Proposal: Each in the range of 1-8

3 Today s Agenda Divya s presentation on the code decay paper Discuss Belady & Lehman s paper Discuss Code Decay paper Q and A session on my feedback to your proposals

4 Today s Presenter Divya Gopinath (Skeptic)

5 A Model of Large Software Development L.A Belady and M.M Lehman 1976 IBM OS/360 Seminal empirical study paper in software evolution

6 What was it like in 1976? IBM PC came out around 1984 Apple introduced PC in 1970s

7 What was it like in 1976? E.W. Dijkstra: a program must as a mathematical theorem should and can be provable Increasing cost of building and maintaining software was alarming

8 Subject Program & Data OS/ years old 20 user-oriented releases Starting with the available data, they attempted to deduce the nature of consecutive releases of OS/360

9 When observing software evolution, what can you measure? # of bugs (reported problems) # of modules => # of directories, # of files performance metrics => times, memory usage, CPU usage, IPC change types: corrective, adaptive, perfective # of developers working in the organization, # of deveopers per module size of code changes => # of lines of changed code how wide spread the changes are => # of files or modules touched by the same change age in calendar year, days, months / age in logical unit such as release, check-in, version

10 Variables observed in Belady and Lehman 1976 The release number Days between releases The size of the system The number of modules added, deleted, and changed Complexity: the fraction of the released system modules that were handled during the course of release MH_r / M_r Manpower, machine time, and costs involved in each release

11 Growth Trends of System Attribute Counts With Time COMPONENTS. TIME MODULES HANDLED

12 Growth Trends Figure 2 Average growth trends of system attributes Figure 3 Average growth trends of system attributes compared with planned growth Figure 4 Serial and average growth trends of a particular attribute REND VERAGE LEGEND: 0 SIZE IN MODULES X MODULES HANDLED +RELEASE INTERVAL I AVERAGE RELEASE SEQUENCE NUMBER I LEGEND: 0 Sl2E IN MODULES X MODULES HANDLED + RELEASE INTERVAL I AVERAGE RELEASE SEQUENCE NUMBER RELEASE SEQUENCE NUMBE

13 What can you deduce from these graphs? It takes longer and longer to release the next release The size of increases over time Figure 3 Average growth trends of system attributes compared with planned growth It requires modifying more and more modules per each release I LEGEND: 0 Sl2E IN MODULES X MODULES HANDLED + RELEASE INTERVAL I AVERAGE RELEASE SEQUENCE NUMBER

14 What can you deduce from these graphs? Figure 4 Serial and average growth trends of a particular attribute Handling fewer number of modules as the software evolves REND VERAGE RELEASE SEQUENCE NUMBE

15 Belady & Lehman: the Law of Program Evolution Dynamics 1. Law of continuing change: a system that is used undergoes continuing change until it is judged more cost effective to freeze and recreate it 2. Law of increasing entropy: the entropy of a system (its unstructuredness) increases with time, unless specific work is executed to maintain or reduce it.

16 Belady & Lehman: the Law of Program Evolution Dynamics 3. Law of statistically smooth growth: Growth trend measures of global system attributes may appear to be stochastic locally in time and space, but statistically, they are cyclically self-regulating, with well-defined longrange trends

17 Belady & Lehman: the Law of Program Evolution Dynamics Law of continuing change: a system that is used undergoes continuing change until it is judged more cost effective to freeze and recreate it!m_r = S1 + Z1 Law of increasing entropy: the entropy of a system (its unstructuredness) increases with time, unless specific work is executed to maintain or reduce it. C_r = R^2 + S2 + Z2 Law of statistically smooth growth: Growth trend measures of global system attributes may appear to be stochastic locally in time and space, but statistically, they are cyclically selfregulating, with well-defined long-range trends M_r = R + S + Z (where S and Z represents cyclic and stochastic components)

18 As a skeptic: Laws are presumptive How do you use for real daily software development? External validity: does the law hold for other projects? Factors:

19 As a skeptic: What is the unit of a module and a component? What is the granularity of a release? Do they have the same amount of functionality addition per each release? What types of changes does each release include? Any changes in the organization structures & developers? Are they laws or just hypotheses? What are potential contributions / benefits of understanding software evolution?

20 My general thoughts on Belady & Lehman Very insightful paper at the time of 1976 The first use of statistical regression for characterizing software evolution Discussed the nature of software evolution, characterized it using their empirical data Deduction of laws from one system s evolution --- very weak external validity, perhaps hasty conclusions

21 Does Code Decay? Eick et al. TSE 2001 (almost 25 years after Belady & Lehman s Study)

22 Problem Definition What do the authors mean by code decay? it is harder to change than it should be related to Belady & Lehman s second law: the entropy of a system increases with time, unless specific work is executed to maintain or reduce it.

23 Discussed Problem Check whether code decay is real: Does Code Really Decay? how code decay can be characterized the extent to which each risk factor matters *Empirical Study* Paper

24 Hypotheses What the authors are trying or expecting to find? The span of files increases over time (age) Effort has some relations to many measurable variables. Modularity breaks over time Fault potential has some relation to many measurable variables.

25 Hypotheses What the authors are trying or expecting to find? 1. The span of changes increases over time 2. Breakdown of modularity increases over time 3. Fault potential, the likelihood of changes to induce faults has some relations to Efforts has some relationship too Usually *good* empirical study paper either finds surprising empirical evidence that contradicts conventional wisdom or provides thorough empirical evidence that validates well known hypotheses.

26 Study Approach Data selection Selection of measurement variables (so called independent variables) Study method that finds *relationships* among the measurement variables

27 Study Approach: (1) Data Selection Rich data set Telephone switching system This system evolved by following a well-defined process. Structured, manually labeled data Easy to group related changes 100 million LOC 5000 modules 50 major subsystems in C and C+

28 Study Approach: (2) Measure Independent Variables c denotes changes (mostly a MR) Variables DELTAS(c) = # of deltas associated with c ADD(c) = # of lines added by c DEL(c) = # of lines deleted by c DATE(c) = the date on which c is completed INT(c) = the interval of c (calendar time required to implement c) DEV(c) = number of developers implementing c

29 Study Approach: (2) Measure Independent Variables Derived variables ˆ FREQ(m, I) = X 1fDATE c 2 Ig; / I c e > m FILES(c) = # of files touched for change c NCSL (m) = # of non-commentary source lines per module AGE(m) = average age of its consequent lines

30 Study Approach: (3) Finding Correlation Linking risk factors to symptoms Statistical regression This requires designing some template models

31 =DEF<1 )=F=G G<HG<I<B?I?;< J:< => J I=EGK< K=D< F9B<1 %J9BL=A K=F=G<D L=M<I G<HG<I<B? >G<NE<B?FO K;JB:<D >9F<I. B< C=DEF<1 )=F=G G<HG<I<B?I?;< J:< => J I=EGK< K=D< F9B<1 %J9BL=A K=F=G<D L=M<I G<HG<I<B? >G<NE<B?FO K;JB:<D >9F<I. <I<B? >9F<I?;J? K;JB:<D F9??F< I9BK<?;<9G KG<J?9=B1 < G<HG<I<B? >9F<I?;J? K;JB:<D F9??F< I9BK<?;<9G KG<J?9=B1 Study Approach: (3) Finding Correlation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ault Potential 1. The! number faults thatq'6;0#) will have to be fixed 5 6;; <0;%& " (%;1"A0"A #1!= 0" *1,';% -6",4 of #$'&4 1< #$% 5%&:1"&%3 0" 6in module m %;6(156#% 0".%/#01" 24 %C#%"&07% <0C%, <'#'5% 0"#%576; 1< #0*%4 #6N%" <51* FKSH4 0"/;',% #$% in the future.0"#%576; Change 1< effects are#6n%" dampened 156#% 0".%/#01" 24 %C#%"&07% #0*%4 <51*over FKSH4time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he number faults that will have to be fixed $%6*!%!,# in a module 6", #$% A%"%56;0@%, ;0"%65 of *1,%; $#1 >, $,3/!&"! $, #+! $!(" 1< 6 &1<#+65% '"0#! 6& Fault #$% Potential 9,! %&4Faults 3(,"#/.3# 6/!'$3#(/" #+!!55(/# :6!/"(,;+(./"0! in 6the6", future. are less likely in(5older code)*$!(" (when beta is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of J=K=J +$%5% )4 +4 $0 4 6", $7 65% $, G+$3+!&3+ 6($,# 3(//!"6(,'" Effort,0&/'&&%, Model: predictors the6", person-hours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iryung!" -J34Kim #$%1#$%5&E!/&=!"- G+!/! #+! G!$=+#" +&E! &:1"#6"%1'&;)4 #$0& "1#Spring #6'#1;1A)O 1< EE382V Evolution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

32 Results: (1) The span of changes increases over time?

33 Results: (2) Breakdown of modularity increases over time? If modules have changed together as a part of the same MR, they were placed to close to each other.

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arge, recent changes %C#%"&07% <'#'5% 0"#%576; 1< #0*%4 #6N%" <51* FKSH4 0"/;',% >36)*$" I1L"$) 8!+ "$) :3 4!-8,&#7 :&) '*"63 )43 '*B3 $, 64!"%3 #$% 53,*"*)*C31 )43*!%3'7 83!'&-35 *" +3!-'H7!"5 )43*'*B3'7!' *" FGHS ;-*8!-+ F!"5 5*-36)H 3C*53"63 )4!) 64!"%3' *"5&63,!&#)'S 6", #$% A%"%56;0@%, ;0"%65 *1,%; "0#! 6& #$% +%0A$#%, #0*%,6*: *1,%; *'add )43*8$-3 5!)! #*93#+,-$8 most to ';!"7 fault *:;%C0#)E *),*)*'&'*"%! '3) " T43-3 ( # $7 ;!') 64!"%3' $, )43 '!83 :3! )*$ 0" $!(#!-2,3$4% '33*"% )43 '*B3 3,,36)!-3 $, ';!"1 )*$!(" $1<&'%# 0" 6 &!+! $1%/"&#,-$8! ' )* 2' "$)35 *"C!-*!:#3 >36)*$"8!'9 L1.7 )43,3!)&-3' )43 & $7 '& $"3!*% +$)%*$.)-$!1"& #!9" '*8;#+!++,*#6!(# potential. *"5*')*"%&*'4!:#3!"$)43!"57 43"637 "$"3 6$&#5 " $!(# /" #$ (!+ #,-. 0% %)8 *1'& +0#$,&"6)*$"!#*)+ $'%!"3 4+'&12 -+% 5--+%*F31%17 :3 ;$'*)35 )$ 4!C3#!"+ ';36*,*6 3,,36)1 & 1< #$% ;0"%& % &!# >S.)-$!1"'/!5"!"5!-3,$-)$$!%%-3%!)35,$- 8$ " 1% ' -3/$4' )' ': #1 #0*%1% *)4!) 6",'$83 )*$ 0& 8 0&8$53# #$% %"#05% &%#(2--$4' 1< ;-$C*53,%;#6& F/H 567 5$3' 3C*53"63 8$5&#3' 3O)3"535!''3'' )43 3C*53"63 W:$))$8 #*"3X -3#3C!"63 K &$1+& #$% +$%5%Q43 Code having many lines A$@3C3-7 3,,$-) 5!)! F;3-'$" 4$&-'H!-3!C $7*''&3,0&/'&&%, 0"536!+35.%/#01" J=K=J 6", +$%5% )4 +4 $0 4 6", ' 234 %)..!1# (.$#!1# (" &65% 6$53 536!+S D!" )43 3,,$-) -3(&*-35 )$ *8;#383") 64!"%3!-3)43 8$-3 =" (" ;-*"6*;#37 )4*' 6$&#5 % ;!-!83)3()4!" #!(#$)43-'1 53)3-8*"35 :+ ')!)*')*6!# )4*',-$8 #3C3#1 F[&-)43!, %&#0*6#%, '&0"A &#6#0&#0/6; 6"6;)&0&= :3 ;-35*6)35 '+8;)$8'!"5a!"!#+'*' -*'9,!6)$-'$,,$-,!6)$-' 536!+\ Q4 :3!55-3''35 :+!##$@*"% ( )$64!"%3' :3 8$5&#3 53;3"53")7 :&) that have survived for %"#06;0#)3 6",!"!#+'*' F'33 KLMNH1 Q4&'7 #!-%37 *1,%; -363")!55 )43 8$') 6", #$% A%"%56;0@%, ;0"%65 T1#$ 1< #$%&% 0",0/%& 0;;'*0"6#% 6& #$% :50*65)!"!#+'*' 38;#$+'! 6& #$% :!'35 $" #!-%3- $, 5!)! :&) -3(&*-*"%! C!-*!") )43 '3)' ;-35*6)*C3 DE= P[[ $,)43 4!C3 "$)!"5 +3) 5$"3 )4*'1 $,/$6"A% /1;15%, ;0"%& )$,!&#) ;$)3")*!# )43 "&8:3)*83'! 8$5&#3 4!' long time is likely to be! 6A%"#2" /5%6#0"A <6';#&=F!"5 -U7%" <6';#&8$53#7,1)*$!(" "1# )43 W'&8,*#3 '*B3'X 64!"%3' )3-8 *" F<H%*C3" $8*))351!%% Q4 3,,$-)$, )$&6435,$- *"5*C*5&!# )*$ 0" 65% 6 :33"!#)3-"!)*C3 #3''#$1'A$ ;$@3-,&#H &'*"% DE= 64!"%35 *'! :3))3;-35*6)$)4!" *)' '*B3 $, )43 5%, ;0"%& &!(# /" # $ '!*%!+ ' + #!9"!+ *#6 #$0& 0& "1#7 6 #6'#1;1A)O"?$% 6(&%"/% 1< 1#$%5-3'&#)'!-3 '&%%3')*C3 :&)7 :36!&'3 $, )43 '8!## '!8;#3 '*B3 &:1"#6"%1'&;)4 relatively free faults. *' of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oftware Evolution: Spring 2009, Instructor Miryung Kim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esults: (3) Fault potential, the likelihood of changes to induce faults increases over time

35 Results: (4) Prediction of efforts increases over time X EFF c ˆ a 0 a 1 FILES c a 2 1fc > e fgjfj a 3 ADD c a 4 DEL c a 5 INT c a 6 DEV c : j j f log 1 EFF c ˆ :32 :13 log 1 FILES c Š 2 :09 log 1 DEL c Š 2 :12 log 1 ADD c Š log 1 DEL c Š :11 log 1 INT c Š :47 log 1 DELTAS c Š: File span has positive correlation. Large deletions are implemented rather easily. Hardest changes require both additions and deletions. Large number of editing changes are rather easy to implement.

36 Expected results? Any unexpected results?... The number of developers does not have much impact Complexity metrics did not play much roles compared to size and number of changes

37 Expected results? File span increases over time. Size and time of changes have some positive correlation with fault potential Modularity degrades over time. Any unexpected results? Once size and time of changes are taken into account, other variables (e.g. # developers, complexity metrics, span of files) did not play much roles in predicting faults. Large number of editing changes are rather easy to implement. In the beginning of evolution, file span was relatively large.

38 Threats to Validity? Limitations?

39 Four types of threats to validity External Validity: Can we generalize to other situations? Internal Validity: Assuming that there is a relationship in this study, is the relationship a causal one? Construction Validity: Assuming that there is a causal relationship in this study, can we claim that the program reflected well our construct of the program and measure? Conclusion Validity: Is there a relationship between the cause and effect?

40 Another way of looking at Validity Theory: what you think Cause Construct cause effect constructs Effect Construct Program programoutcome Observations Observation: what you test

41 Example: WWW => Student Learning Theory: WWW virtual classroom improves student understanding of course materials WWW virtual classroom cause effect constructs Understanding Second life instruction programoutcome Test score Observation: Let one half of EE382V to use second-life virtual class room and let the other half come to regular lecture. Compare their test scores at the end.

42 External Validity Theory: WWW virtual classroom improves student understanding of course WWW virtual classroom cause effect constructs Understanding Second life instruction programoutcome Test score Observation: Let one half of EE382V to use www site and let the other half External Validity: Does this study generalize to students of EE322c?

43 Internal Validity Theory: WWW virtual classroom improves student understanding of course WWW virtual classroom cause effect constructs Understanding Second life instruction programoutcome Test score Observation: Let one half of EE382V to use www site and let the other half Internal Validity: Assuming that students using WWW did better in their test, isn t it because these students have more money (apparently they have computers & high-speed internet) and rich students have more experiences with objective tests (due to their parents sending them to prep-schools.)

44 Construct Validity Theory: WWW virtual classroom improves student understanding of course WWW virtual classroom cause effect constructs Understanding Second life instruction programoutcome Test score Observation: Let one half of EE382V to use www site and let the other half Construct Validity: Is the operationalization method valid? Do objective test scores truly reflect students understanding of core concepts? Don t students who are familiar with second life interface just test better?

45 Conclusion Validity Theory: WWW virtual classroom improves student understanding of course WWW virtual classroom cause effect constructs Understanding Second life instruction programoutcome Test score Observation: Let one half of EE382V to use www site and let the other half Conclusion Validity: Are the correlation between second-life virtual classroom use and test scores significant?

46 1. Temporal Behavior of the Span of Code Changes Cause Construct cause effect constructs Effect Construct Program programoutcome Observations

47 1. Temporal Behavior of the Span of Code Changes External Validity Internal Validity Construct Validity Conclusion Validity

48 2. Time Behavior of Modularity Cause Construct cause effect constructs Effect Construct Program programoutcome Observations

49 2. Time Behavior of Modularity External Validity Internal Validity Construct Validity Conclusion Validity

50 3. Prediction of Faults Cause Construct cause effect constructs Effect Construct Program programoutcome Observations

51 3. Prediction of Faults External Validity Internal Validity Construct Validity Conclusion Validity

52 4. Models of Effort Cause Construct cause effect constructs Effect Construct Program programoutcome Observations

53 4. Models of Effort External Validity Internal Validity Construct Validity Conclusion Validity

54 My general thoughts on Code Decay Paper Rich data set!!! Scientific research method Identification of hypotheses => identify key variables and measure them = > create statistical models => statistical regression What do identified coefficients real mean? Can programmers use any of these findings for daily development activities?

55 Recap Code decay can be mapped to specific measured / derived variables. e.g., span of changes => file span, non-localized changes => changes that spans module boundaries Early mining software repositories research in late 90s that is based on statistical regression analysis and visualization These types of research require having good insights. e.g., weighted time dampened model Identified which factors do mater! => some surprising results that complexity metrics do not matter

56 Limitations Carefully designed model that may have over-fitted data? Their model did not consider change types or change content Their model cannot handle module specific information Their results do not generalize to other systems because most changes in open source system does not map to a logical software change