Code Beauty

Prof. Fabio Kon Department of Computer Science Institute of Mathematics and Statistics University of So Paulo SugarLoafPLoP'2010 - 24/9/2010 Salvador, Brazil

Agenda

Definition of Beauty Is Code Beauty important for Software Development? What is Code Beauty

Beauty - Ancient View

pre-socratic (e.g., Pythagoras): strong connection among Beauty and Mathematics; objects with proportions following the golden ratio are more attractive to the human brain
two quantities are in the golden ratio if the ratio of the sum of the quantities to the larger quantity is equal to the ratio of the larger quantity to the smaller one. The golden ratio is an irrational mathematical constant, approximately 1.6180339887

greek architecture is based on symmetry and on the golden ratio

3

Beauty - Romantic View

works do not need to be harmoniously proportioned and constitute a well-rounded whole to be beautiful; on the contrary, dissonance and fragmentation is beautiful more precisely: dissonance and fragmentation allude to a harmony and wholeness that is not in the work
4

Beauty - Modernist View

architect Mies van der Rohe: Less is More

~ 1940

designer Buckminster Fuller (geodesic dome) adopted the engineer's goal of Doing more with less writer Antoine de Saint Exupry (1939) perfection is attained not when there is nothing more to add, but when there is nothing more to remove
5

Beauty: Post-modernist view

architect Robert Venturi: Less is a Bore, let's do whatever people like

the greek and modernist views of beauty provide us with a good basis for approaching Beauty in Science and Technology

Science
make everything as simple as possible, but not simpler

Technology
Conceptual integrity

the architect or team of architects should develop an idea of what the system should do and make sure this vision is understood by the rest of the team to ensure a user-friendly system, a system may deliberately provide fewer features than it is capable of

Beauty dictionary definition

the quality that gives pleasure to the mind or senses and is associated with such properties as

harmony of form or color, excellence of artistry, truthfulness, and originality.

The American Heritage Dictionary of the English Language, 4th Edition, 2000.
10

Does it matter?
Why are we talking about Code Beauty on a serious Software Conference???

personally, I have two reasons...

11

there has been a strong movement towards managing software development projects as any other production process/system many cases of managers that have never written a single line of code is it possible to have a good Chef de Cuisine that has never cooked in his life?
12

scientists/engineers/practitioners sometimes tend to overestimate the power of science and mathematics

can science and engineering provide all what's needed to explain life? and software development?

13

What is Software Development?

Modeling Engineering Discipline Poetry Craft Art

(Jacobsen) (Meyer) (Humphreys) (Cockburn) (Knuth) (Gabriel)

(from from Alistair Cockburn)

Common mistake: look at software as only one of the above items.

14

our goal here, is to bring attention to a few important aspects of software development that are neglected in some software engineering communities

we'll look at the most important deliverable on a software project:

The Code
15

Beauty is fundamental
vinicius de morais

Beautiful code

brings pleasure to the reader makes the writer happy makes working in groups fun fewer bugs maintainability team productivity in other words, Quality
16

This leads to

In Software Development, Beauty Leads to Quality

17

What is Beauty in Software Development?

sources:

Robert C. Martin. Clean Code - A Handbook of Agile Software Craftsmanship. Prentice Hall. 2008. Andy Oram and Greg Wilson. Beautiful Code. O'Reilly. 2007. beaute(code) art exhibit by Bob Hanmer, Karen Hanmer, and Andrea Polli. exchange with experts
18

Rebecca Wirfs-Brock's code beauty

it is a common recognition that these idioms and those structures and those ways of doing things are of value

it is important that a project or a team have the same sense of aesthetics or there will be clashes of will

keeping of every step in a method at the same level of abstraction/intention; it reads much more like prose that way

characteristics that are "normally" present in beautiful code are balance, effectiveness, expressiveness, and profoundly doing well what it was designed to do.

19

the Smalltalk class collection

Rebecca's example of a beautiful code

provide all the built in behavior by requiring subclasses to implement three methods:

add: anObject remove: anObject ifAbsent: exceptionBlock do: aBlock

methods in the abstract class collection to add, remove, check for emptiness, inclusion, occurrences, and to step through a collection in different ways accumulating values are all implemented use these abstract building blocks or invoking other defined behaviors that depend on these implementation; elegant, beautiful reuse!

20

Joe Yoder's code beauty

I think it is important to most of us who live there; it is like cleaning up our living area, having clean code or beautiful code can make life much better

however let's face it, BBoM still thrive and exist and are successful

it matters, but is not necessary and sufficient for successful software projects. If so, then Clean Code would become the norm

those of us who care strive for it because we believe it is the right thing to do and it will help us in the long run...but certain forces are in play that can make clean code or beautiful code a goal that is very hard to achieve

21

Joe's example of a beautiful code

also the Smalltalk class collection you see a certain beauty and elegance in the system as it was easy to understand and extend

(it's also my favorite piece of code :-)

design patterns can help with this as well; if a system is clean and well designed by implementing some of the best known principles and practices one can see certain beauty in the system

22

W h a tis C l a n C ode ? e

Bjarne Stroustrup
Inventor of C++

I l m y co de t be e l e g a n t a nd ike o e ffi c i e n t. Th e l gic sh o ul be o d s t ra i g h forw a rd t m a ke it h a rd o fo r bugs t o h ide , te h dependenc i es mi ni mal to ease mai ntenanc e, e rror h a n d l i n g c om p l e t e a cco rding t o an a rt a t d icul e st t gy, ra e a nd p e rform a n c e c l os e to op t i m a l so a s not t t m p t p e op l t m a ke o e e o t e co de m e ssy w it unp rincip l d h h e op t iza t im ions. Cl e a n c od e d oe s on e t h i n g w e l l .

23

W h a tis C l a n C ode ? e

C l a n co de is s i m p l e a nd d i re c t. e C l a n co de re a d s l i k e w e l l e w ri t te n p ros e . C l a n code ne ve r e obscure s t e de signe r's int nt but h e ra t e r is ful o f crisp [cl a rl de fine d] h l e y a bst ct ns a nd s tra i g h forw a rd ra io l s o f co nt l ine ro .

Grady Booch
Author of Object Oriented Analysis and Design with Applications

24

W h a tis C l a n C ode ? e

Dave Thomas
Founder of OTI, godfather of the Eclipse Strategy

Cl e a n c od e c a n b e re ad , a nd e nh a nce d by a de ve l p e r o t e r t a n o h h it origina l a ut or I t h a s u n i t a n d s h . ac c eptanc e tests. It h a s m e a n i n g fu l n am e s . It p rovide s on e w a y ra t e r t a n m a ny w a ys h h fo r do ing o ne t ing. It h a s m i n i m a l h dependenc i es, w h ich a re e xp l l de fine d, a nd p ro vide s a icity cl a r a nd m i n i m a l AP I . C ode e sh o ul be l i te ra t e since de p e nding d on t e l ngua ge , not a l ne ce ssa ry h a l info rm a t n ca n be e xp re sse d io cl a rl in code a l ne . e y o

25

W h a tis C l a n C ode ? e
I co ul l a l o f t e qua l ie s t a t I d ist l h it h no t ice in cl a n code , but t e re is e h one ove ra rch ing qua l y t a t l a ds t it h e o a l o f t e m . Cl e a n c od e a l w a ys l h l ook s i t w a s w ri t t e n b y s om e on e w h o c a re s . Th e re i s n ot h i n g obvi ou s t h a t you c a n d o t o m a k e i t b e tt e r. A l of l t o se t ings w e re t o ugh t a bo ut by h h h t e co de 's a ut or a nd if you t t h h , ry o im a gine im p ro ve m e nt yo u're l d s, e ba ck t w h e re you a re , siting in o t a p p re cia t n o f t e code som e one io h l ft fo r yo u c od e l e ft b y e s om e on e w h o c are s d e e p l y a b ou t th e c ra ft.

Michael Feathers
Author of Working Effectively With Legacy Code

26

W h a tis C l a n C ode ? e

In re ce ntye a rs I be gin, a nd ne a rl y e nd, w it B e ck's rul s of sim p l h e e co de . In p rio rit o rde r, sim p l code : y e R u n s al l t e s t s Con ta i n s n o d u p l i c a t i on E x p re s s e s al l t h e d e s i g n i d e a s t a ta re in t e syst m h h e Mi n i m i ze s th e n u m b e r of e n t i ti e s such a s cl sse s, m e t o ds, a h funct ions, a nd t e l . h ike

Ron Jeffries
Author of Extreme Programming Installed

27

W h a tis C l a n C ode ? e

Y o u kno w you a re w orking on cl a n e co de w h e n e a c h rou ti n e you re a d tu rn s ou t t o be p re t t y m u c h w h at you e x p e c t e d . Y o u ca n ca l it b e a u t i fu l code w h e n t e l h co de s a l m a k e s i t l ook l i k e so t h e l a n g u a g e w a s m a d e for t h e p rob l e m .

Ward Cunningham
Inventor of Wiki, Fit and much more Godfather of all those who care about code

28

W h a tis C l a n C ode ? e
S im p l e W it outo bvio us h im p rove m e nt s

E fficie nt S t igh t rw a rd ra fo

E xp re ssive C o nt ins no a dup l t ica ions

Turns outt be w h a t o you e xp e ct d e F ul o f m e a ning l

R uns a l t st le s
L it ra l e

R e a ds w e l l M inim a l

B e a u t i fu l : when th e l anguage was made for th e p rob l e m

W rite n by t som e one w h o ca re s

29

language may influence

/* C Language */ #include <stdio.h> #define NumLines 10 /* number of lines */ main(argc,argv) int argc; char **argv; { int tri[NumLines][NumLines], /* triangle */ r,c,i; /* row, column, misc index */ /* initialize array */ for (r=0;r<NumLines;r++) { tri[r][0]=1; for (c=1;c<NumLines;c++) { tri[r][c]=0; }; /* generate triangle */ tri[0][0]=1; for (r=1;r<NumLines;r++){ for (c=1;c<NumLines;c++) { tri[r][c]=tri[r-1][c] + tri[r-1][c-1]; } }; /* print triangle */ for (r=0;r<NumLines;r++){ for (i=0;i < ((NumLines-r-1)/2);i++) printf(" if (r%2 == 0) printf(" "); for (c=0;c<NumLines;c++){ if (tri[r][c] != 0) printf("%3d ",tri[r][c]); }; printf("\n"); }

");

source: http://www.andreapolli.com/beaute%28code%29/good_bad_ugly/pascal.html

30

let's get a little more technical

31

Me an i n g fu l Names

32

M e a ningful N a m e s

N a m e s are vi t a l !

C ode is ba sica ly na m e s a nd re se rve d w o rds l C h o o sing go od na m e s t ke s t e butsa ve s a im

m o re t a n itt ke s h a N a m e s sh oul be e xp re ssive a nd sh o ul d d

a nsw e r que st ns io

33

M e a ningful N a m e s

E xampl e
public List<int[]> getThem() { List<int[]> list1 = new ArrayList<int[]>(); for (int[] x : theList) if(x[0] == 4) list1.add(x); return list1; }

34

M e a ningful N a m e s

E xampl e
public List<int[]> getThem() { List<int[]> list1 = new ArrayList<int[]>(); for (int[] x : theList) if(x[0] == 4) list1.add(x); return list1; }

M a ny do ubt a rise . . . s 1 . W h a tdoe s t is m e t od ge t h h ? 2 . W h a tkinds o f t ings a re in t e L ist h h ? 3. W h a tis t e im p ort nce of t e ze rot p osit h a h h ion? 4. W h a tis t e significa nce o f t e va l 4? h h ue
35

M e a ningful N a m e s

E xampl e
public List<int[]> getThem() { List<int[]> list1 = new ArrayList<int[]>(); for (int[] x : theList) if(x[0] == 4) list1.add(x); return list1; }

W h a ta bo utt is h code ?
public List<int[]> getFlaggedCells() { List<int[]> flaggedCells = new ArrayList<int[]>(); for (int[] cell : gameBoard) if(cell[STATUS_VALUE] == FLAGGED) flaggedCells.add(cell); return flaggedCells; }
36

M e a ningful N a m e s

E xampl e
public List<int[]> getFlaggedCells() { List<int[]> flaggedCells = new ArrayList<int[]>(); for (int[] cell : gameBoard) if(cell[STATUS_VALUE] == FLAGGED) flaggedCells.add(cell); return flaggedCells; }

P robl m e sol d! ve
1 . W h a tdo e s t is m e t o d ge t Itge t a l fl gge d ce ls! h h ? s l a l 2 . W h a tkinds o f t ings a re in t e L ist t e L istis a ga m e B oa rd file d w it h h ? h l h ce ls! l 3. W h a tis t e im p o rt nce of t e ze rot p osit n? Th a t t e S t t Va l ! h a h h io 's h a us ue 4. W h a tis t e significa nce o f t e va l 4? Itm e a ns itis fl gge d! h h ue a
37

M e a ningful N a m e s

E xampl e
public List<int[]> getFlaggedCells() { List<int[]> flaggedCells = new ArrayList<int[]>(); for (int[] cell : gameBoard) if(cell[STATUS_VALUE] == FLAGGED) flaggedCells.add(cell); return flaggedCells; }

G oing furt e r . . h .
public List<Cell> getFlaggedCells() { List<Cell> flaggedCells = new ArrayList<Cell>(); for (Cell cell : gameBoard) if(cell.isFlagged()) flaggedCells.add(cell); Th i s i s p re t ty return flaggedCells; } muc h wh at

you e x p e c te d !

38

Ch a n g e s s h ou l d b e e as y!

Us e R e a d ab l e n am e s

To m a ke ch a nge s, w e ne e d t unde rst nd t e o a h co de !

X Y Z C o nt l rH a ndl role ingO fS t rings != X Y Z C ont le rS t ge O fS t ro l ora rings

Us e Se a rc h a b l e n am e s Us e t h e L an g u ag e St an d a rd s Us e Sol u t i on Dom a i n n a m e s

U se p a te rn a nd a l t gorit m na m e s, m a t t rm s, h h e

Us e P rob l e m Dom a i n n a m e s Don 't c on fu s e t h e re a d e r

U se t e O ne W ord P e r C o nce p t rul h e D o n'tuse j s, m ind m a p p ings, h unga ria n no t t oke a ion, . . .

39

M e a ningful N a m e s

Con c l u s i on

itis e a sy t sa y t a tna m e s sh oul re ve a l int nt W h a t o h d e .

w e w a ntt im p re ss up o n yo u is t a tw e a re s e rio us a bo ut o h t is h

sh o rtna m e s a re ge ne ra ly be te r t a n l nge r o ne s, so l t h o

l ng a s t e y a re cl a r (do notp ro m o t obscurit t sa ve a o h e e y o co up l o f ke yst ke s) e ro

i f you fi n d a b a d n a m e , c h a n g e i t n ow !

40

Fu n c t i on s

41

F unct ions

Fu n c ti on s s h ou l d b e smal l !

fu n c t i on s s h ou l d h a ve fe w l i n e s

e a ch of t e m sh oul be o bvio us a nd e a sy t unde rst nd h d o a

fu n c t i on s s h ou l d n ot h ol d n e s te d s t ru c t u re s

if, w h il , e l bl cks sh o ul be st igh t e se o d ra forw a rd (p roba bl a funct ca l) y ion l t e condit na l sh oul p roba bl be a funct n ca l t a te nca p sul t s it h io d y io lh a e

42

F unct ions

On e Th i n g !
F unct ions sh oul do one t ing. d h Th e y sh oul do itw e l. d l Th e y sh o ul do itonl d y.

fu n c ti on s t h a t d o on e t h i n g c an 't b e d i vi d e d i n t o s e c t i on s t w o w a ys t o i d e n t i fy w h e th e r a fu n c t i on d oe s On e Th i n g

if a funct do e s onl t ose st p s t a ta re one l ve l be l w t e st t d ion y h e h e o h a e na m e of t e funct h ions, t e n t e funct ns is do ing one t ing h h io h If you ca n'te xt cta no t e r funct n from itw it a na m e t a tis no t ra h io h h m e re l a re st t m e ntof it im p l m e nt t y a e s e a ion, t e n it doing o ne t ing h 's h

43

E xampl e
public void pay() { for (Employee e : employees) { if (e.isPayday()) { Money pay = e.calculatePay(); e.deliverPay(pay); } } }

Itdoe s m ore t a n one t ing. . . h h

1 . itl op s ove r a l t e e m p l e s o lh oye 2 . ch e cks t se e w h e t e r e a ch e m p l e ough tt be o h oye o p a ye d 3. p a ys t e e m p l ye e h o

44

R e fa c t ore d E xampl e
public void pay() { for (Employee e : employees) payIfNecessary(e); } private void payIfNecessary(Employee e) { if (e.isPayday()) calculateAndDeliverPay(); } private void calculateAndDeliverPay(Employee e) { Money pay = e.calculatePay(); e.deliverPay(pay); }

Itj it ra t s ove r ust e e t e em pl es h oye

C h e cks w h e t e r a n h e m p l e o ugh tt be p a id oye o

P a ys t e h e m p l ye e o

is this exaggerated?
45

F unct ions

On e l e ve l of Ab s t ra c t i on

s t a t e m e n t s w i th i n a fu n c t i on s h ou l d b e a l l i n th e s a m e l e ve l

m i x i n g l e ve l s i s c on fu s i n g

o nce de t il a re m ixe d w it e sse nt l conce p t m ore a nd m ore a s h ia s, de t il t nd t a ccre t w it in t e funct a s e o e h h ions it t e firstst p t a rds t e cre a t of big funct 's h e ow h ion ions!

t h e St e p d ow n ru l e

w e w a ntcode t re a d l a t -do w n na rra t o ike op ive

46

F unct ions

Fu n c ti on a rg u m e n t s

fu n c ti on s s h ou l d m i n i m i ze t h e n u m b e r of a rg u m e n ts

a rgum e nt a re h a rd from a t st p o into f vie w s e ing t o m a ny a rgum e nt = t e funct doe s m o re t a n o ne t ing o s h ion h h t o m a ny a rgum e nt = t e funct is use d in m a ny diffe re ntw a ys o s h ion

d on 't u s e fl ag arg u m e n t s

itl oudl p ro cl im s t a tt e funct y a h h ions is doing m o re t a n one t ing h h

47

F unct ions

L as t b u t n ot l e a s t

fu n c ti on s s h ou l d n ot h a ve s i d e e ffe c t s

t e y usua ly cre a t t m p ora l co up l h l e e ing itsh o ul cre a t a e ffe cton t e obj ctor re t so m e t ing d e h e urn h

Don 't R e p e a t You rs e l f (DR Y)

dup l t n m a y be t e rooto f a l e vil in soft a re ica io h l w

funct ions sh o ul be sh ort w e l na m e d a nd nice l orga nize d d , l y

48

Com m e n t s

N o t ing ca n be quit so h e l ful a s a w e l-p l ce d com m e nt h e p l a . N o t ing ca n cl t r up a m odul t a n frivo l h ute e h ous do gm a t ic co m m e nt N ot ing ca n be quit so da m a ging a s a n ol cruft s. h e d y co m m e ntt a tp rop a ga t s l s a nd m isinfo rm a t h e ie ion.

49

C o m m e nt s

Th e p rob l e m w i th c om m e n t s

c om m e n t s are , a t b e s t , n e c e s s ary e vi l

t e p ro p e r use o f com m e nt is t co m p e nsa t for our fa il h s o e ure t o e xp re ss o urse l in code ; no t t a tI use d t e w ord failure; I m e a ntit f e h h

th ey l i e!

p ro gra m m e rs ca n'tre a l ica ly m a int in t e m ist l a h co m m e nt do n'ta l a ys folow t e code ch a nge s s w l h t e y re quire a m a int na nce e ffo rtt a tt ke s t e h e h a im t h ca n o nl be found in o ne p l ce : t e code rut y a h

50

C o m m e nt s

Cod e i s t h e on l y tru th

51

C o m m e nt s

Good Com m e n ts

l e g a l c om m e n t s in form a ti ve c om m e n t s

C o m m e nt re gul r e xp re ssions ca n be quit use ful ing a e

e x p l an at i on of i n t e n t an d c l a ri fi c at i on s

som e de cisions a re n'tim p l m e nt t de cisions e a ion w e h a ve t use l rie s t a ta re n'tso e xp re ssive o ibra h

Am p l i fi c a ti on

e xp l in h ow im p o rt nta n e l m e ntis a a e

TODO Com m e n t s an d J avad oc s i n P u b l i c AP I

52

53

why should we write beautiful code?

to feel good with ourselves

54

Bibliography

Wikipedia definition for Beauty, Golden Ratio, Minimalism. 2010. Fred Brooks. The Mythical Man-Month: Essays on Software Engineering. 1975. Dorthe Jrgensen. The Metamorphosis of Beauty. 2002. The American Heritage Dictionary of the English Language, 4th Edition, 2000.

55

Bibliography

Robert C. Martin. Clean Code - A Handbook of Agile Software Craftsmanship. Prentice Hall. 2008. Andy Oram and Greg Wilson. Beautiful Code. O'Reilly. 2007. beaute(code) art exhibit by Bob Hanmer, Karen Hanmer, and Andrea Polli. Some material available at http://karenhanmer.com/gallery/?gallery=beautecode and http://www.andreapolli.com/beaute(code) Interviews with experts slides by Joo Machini de Miranda - IME/USP
56