Coding Guidelines

How to not shoot yourself in the foot

Richard Peschke | Coding Guidelines | | Page 1 0. Use the right Tools >Use version control! >Use an IDE! >Use a debugger!

Richard Peschke | Coding Guidelines | | Page 4 General Things

>C++ is most likely the worst language to start programming >We have not chosen it because it is such a nice language >If you are new to >programming start with: . Python, JavaScript or Matlab . a scripting language! . If you are teaching C++ >“STOP TEACHING C”

Richard Peschke | Coding Guidelines | | Page 5 Performance

>Why C++? Because Performance! >Cell phones: How to save energy? Performance! >Computer farm: How to save money? Performance!

>1% of performance boost saves Facebook 100k€ in electricity Source: Ima Dethi-Sup

Richard Peschke | Coding Guidelines | | Page 6 Performance

>Don’t use new >Don’t use Objects features . Objects have the overhead . Fortran is the peak of constructing and deleting of computer science >Don’t use garbage evolution collection >Functions are evil . Modern computer have . Copy all code in enough memory one function to >Exceptions are avoid jumps dangerous >Comments . Don’t write errors . Every three lines

Richard Peschke | Coding Guidelines | | Page 7 Stop doing Cargo Cult Programming

Richard Peschke | Coding Guidelines | | Page 8 Why do I care so little about Performance?

>For the next two years All your programs are just “Hello World” programs They do not need to run fast! Understanding how they work is the most important thing

You are programming for the trash bin! (90% of the time)

Richard Peschke | Coding Guidelines | | Page 9 How do you get Performance?

>By not using your code! >If the performance has Use libraries to come from your code Copy examples from the internet >If you have a problem, somebody else had it before you!

Richard Peschke | Coding Guidelines | | Page 10 Refactoring

Requirements >Readability >Use an environment where you can easily jump through your code . Two button presses to get to the definition of something is to much . Creating function definitions from declaration should not be more than two buttons . The environment needs to work with symbols and not with Text . Do not use regex to search in your code >Automated Build system . Getting to the line where an error occurred must not be more expensive than one click

Richard Peschke | Coding Guidelines | | Page 11 Readability

Understanding your Code

Refactoring Modular Design Documentation Testability

Reuse Sharing Correctness

Richard Peschke | Coding Guidelines | | Page 12 Performance How to get your code nice and clean

>We are no professionals we are just physicists! >Nobody writes bad code on purpose! >Nobody writes good code alone!

Richard Peschke | Coding Guidelines | | Page 13 How to measure code Quality?

Industry standard: >someValue wtf/min >HEP: >100*someValue wtf/min >HEP solution:  don’t have code reviews at all

Richard Peschke | Coding Guidelines | | Page 14 How to start

Before you start the computer >Write down what your code needs to do (in English or your native language) >Find specific tasks >Find the dependencies . (tasks, data, handles, hardware …) . What belongs together >Speak with somebody >Somebody else!

Richard Peschke | Coding Guidelines | | Page 15 What is Programming?

Debugging Documenting

Planning

Testing Coding

Code Review

Richard Peschke | Coding Guidelines | | Page 16 Organizational Stuff

Organizational: >Use a version control system Git SVN >Use automatic build system . Cmake . make . MSBuild

Richard Peschke | Coding Guidelines | | Page 17 Git

Richard Peschke | Coding Guidelines | | Page 18 CMake

Richard Peschke | Coding Guidelines | | Page 19 Design Styles

Keep it Simple! >Code must be human readable! . Code that only you can read is wrong! Even if it gives reasonable results! >Correct is better than fast! >Use self-explaining names >One entity has one purpose . Every function should do exactly one thing Stay on one Level of abstraction . Do not reuse variables . Give classes specific tasks

Richard Peschke | Coding Guidelines | | Page 20 Easy things to avoid

>Copy and paste >Ambiguous names . If you need it twice make it a . Variable names do not need to be short function . Auto completion was invented in 1988 > >Magic numbers Raw Pointer . . Make it a variable, give it a Use Shared_ptr . meaningful name! Do you really need a pointer? > >Global variables Don’t use c style cast . Use dynamic_cast or static_cast . You never need them >C style arrays . Use vector! >Avoid to write loops . Use algorithms

Richard Peschke | Coding Guidelines | | Page 21 Comments

Comments can be gossip

Richard Peschke | Coding Guidelines | | Page 22 Comments

Bad >Don’t rewrite your code in pseudo code in comments . Expect that who ever reads your code knows C++ better than you (maybe even better than English)

Richard Peschke | Coding Guidelines | | Page 23 Comments

Void f1(){ … 1000 lines of code Bad /* Check to see if the employee is eligible for full benefits*/ >You want to say something if((employee.flags & HOURLY_FLAG) && in English what you can’t (employee.age > 65)) say in C++ } Learn programming >Your code is a mess and Do it like this: if(employee.isEligibleForFullBenefits( )) you want to give some hints what it does Learn programming

Richard Peschke | Coding Guidelines | | Page 24 Comments

Void MyMonsterFunctionWithMoreThan1000lines(){ // declaring some parameter // which i need later for sure Bad ... 100 lines // Initilizing Variables >You want to say something ... 100 lines // using Variable in English what you can’t ... 100 lines say in C++ //clean up } Learn programming >Your code is a mess and Do it like this: you want to give some hints Void MyFunctionWithAUsefulName(){ what it does  Object myObjectWithAMeaningfullName; Learn programming myObjectWithAMeaningfullName.Init();

useMyObjectForASpecificTask( myObjectWithAMeaningfullName) maybeDoSomethingElseWithIt( myObjectWithAMeaningfullName)

} Richard Peschke | Coding Guidelines | | Page 25 Comments

Good /*True if collection is the default collection for the given type. This implies that the collection is >At the library, program, or complete and unambigous. Convenient method function level, describe that checks bit BITDefault of the flag word Implements EVENT::LCCollection.*/ what bool isDefault ( ); >Inside the library, program, /* We're going to use newton's method to find the root of a number because there is no analytical or function, describe how way to solve these equations */ >At the statement level,

/* We need to divide items by 2 here because describe why they are bought in pairs*/ cost = items / 2 * storePrice;

Richard Peschke | Coding Guidelines | | Page 26 Magic Numbers

What are magic numbers? Why they are evil? >Plain numbers that appear >You can’t search for them! somewhere in your source >You can’t make changes code >They contain unknown

> HANDLE hSerial; information > hSerial = reateFile("COM1", > GENERIC_READ | GENERIC_WRITE, > 0, > 0, > OPEN_EXISTING, > FILE_ATTRIBUTE_NORMAL, 0); >What do all the zeroes mean?

Richard Peschke | Coding Guidelines | | Page 27 How to avoid magic numbers

On library level >If your function only accepts certain parameters Provide them alongside with the function >Think about using enums

Richard Peschke | Coding Guidelines | | Page 28 How to avoid magic numbers inside a library or a program >Why is it different from global variables? Because they are const! >Use const values over preprocessor definitions . They are type safe >Group them in namespaces >Comment them where you define them!

Richard Peschke | Coding Guidelines | | Page 29 How to avoid magic numbers

Inside a function

… >Your are using the wrong class or type These are >Use specific types for magic Numbers specific tasks

…

Richard Peschke | Coding Guidelines | | Page 30 Identifying Magic Numbers

M agi f1(105);c Nu RAW Strings are Magic numbers too! mbe r >They are not checked by the f1({ 105,32,97, compiler r >Hard to search for 109,32,100,be um >Hard to maintain N ic117,109,98,0}); >Almost impossible to change ag M afterwards f1((char*)new int[] { 105,32,97, 109,32,100,

117,109,98,0});r be um N gic f1("i amM astupid");

mber Magic Nu

Richard Peschke | Coding Guidelines | | Page 31 Magic Numbers (Real Live Experience)

e r m yTTree-> D raw ( e h "m yAxisW ithRidiculousLongN am e w y : r m yO therAxisW ithRidiculousLongN am e e v > > e m yFancyO utputHistogram ", s r "M ySuperAw esom eCutVariable > 5 e & & b M ySuperAw esom eCutVariable < 7“ m ,"CO LZ“ u n ) ic g a M

Richard Peschke | Coding Guidelines | | Page 32 Where is the problem?

>TTrees are generic container The types and names are not provided by the library >Getting objects by there names ( as string) is a common approach for any generic container (zB. file I/O) >Dynamic typed languages overcome the problem by creating during runtime a handler object

N o t p o ss ib le in C + +

Richard Peschke | Coding Guidelines | | Page 33 Magic Numbers (Real Live Experience)

Create your own custom class class M yCustom Tree { public: No Magic (numbers) TAxis_t G etAx1(); Anymore TAxis_t G etAx2(); private: TTree m _tree; }; TAxis_2D _t operator& (TAxis_t, TAxis_t); TAxis_2D _w ith_h2_t operator>> (TAxis_2D _t, const TH 2& ); Long_t D raw (TD raw Axis); int m ain(){ M yCustum Tree m yTree; TH 2* h2 = new TH 2D (… ); D raw ((m yTree.G etAx1()& m yTree.G etAx2())> > *h2); return 0; } Magic is heresy

Richard Peschke | Coding Guidelines | | Page 34 The Three Base Types in C++

>Float (double…) >Insanity (aka unsigned) Use: Use: . To store quantities which can warry . If you need to express your insanity in over many orders of magnitude source code OR: . If you are lazy . Default Type for beginners > Int (long int …) Use: . You need a counter . When you know during compile time what range of values to expect . When Performance really mattes to you (experts stuff)

Richard Peschke | Coding Guidelines | | Page 35 Types and why do we care

void add_date(int, int, int); >How do I use this thing? add_date(10, 11, 12); >What date is it? . 10.11.2012? . 10/11/2012? . 2010-11-12?

Richard Peschke | Coding Guidelines | | Page 36 Types and why do we care

void add_date(Day_t, Month_t, Year_t); add_dateCom (10, 11, 12); piler E add_date(Month_trror (10), Day_t(11), Year_t(12)); Compile add_date(Day_t (10), Month_tr Er(11),ror Year_t(12)); add_dateCompiling(10_dd, 11_mm,Working 12_yy); Doing the right thing

User defined literals:

Richard Peschke | Coding Guidelines | | Page 37 Types: Summery

>Treat build in types (int/double/string) as Use the compiler radioactive . They don’t carry any information >Use typedefs (renaming of types) to show an intention >Use custom container classes to have the compiler check your code

Richard Peschke | Coding Guidelines | | Page 38 Give a class one specific task

In the real world objects are build of smaller object. Why not in your code?

Richard Peschke | Coding Guidelines | | Page 39 examples

Richard Peschke | Coding Guidelines | | Page 40 Use Modern Code!

H ow h ard ca n Example: “for loop” it b e? Vector vec; >Is this snipped correct? … Variable name has no meaning >How does modern code Use of build in type which looks like? has no connection to the containerfor(int i=0;i< vecSigned.size unsigned();i++){ mismatch > for(auto& e_of_vec : vec){ cout< cout<< e_of_vec << “, “ ; } }

cout<

Richard Peschke | Coding Guidelines | | Page 41 RAII

Richard Peschke | Coding Guidelines | | Page 42 RAW Pointer

There might be a leak somewhere

Richard Peschke | Coding Guidelines | | Page 43 Responsibility Resource Management

What are resources? what do you do with them >Memory >Require them >Ports >Work with them >Files >Release them >…

>Coping what. your you computer can’t hasdo: a limited amount of memory (you can copy the elements in memory but not the memory)

Richard Peschke | Coding Guidelines | | Page 44 Example code: Memory Allocation D D Staticallyo Dynamically Variable number of n o elements (good) Fixed number of elements voidn f2(int elem ents){ void’ f1(){ (bad) t int *’i= new int[elem ents]; int i[100]; … t w Nobody knows where it ends … for (autow e= i;e!= i+ elem ents(bad);++ e) r { r for (autoit & e:i) cout< < *e< < endl; e it { } e delete[] i; t cout<

Use a container class Variable number of elements void f3(int elem ents){ (good) vector< int> i(elem ents); … for (auto& e:i) { It knows its own size cout < < e < < endl; (good) } }

Automatic clean up (very good)

Richard Peschke | Coding Guidelines | | Page 46 How does it work?

class Vector{ public: Vector(int Num berOfElem ents){ m _size_of_array= Num berOfElem ents; >Dynamic memory m _pointer= new int[m _size_of_array]; } allocation inside the int* at(int elem ent){ object return m _pointer+ elem ent; } >Provides access to the int size(){ Array return m _size_of_array; } >The destructor calls the ~ Vector(){ delete operator delete[] m _pointer; } >The object store the private: pointer to the array and int* m _pointer; int m _size_of_array; its size };

Richard Peschke | Coding Guidelines | | Page 47 Responsibility Resource Acquisition Is Initialization RAII

Example of an RAII class: Why should you do it? >You can’t forget to release the resource >Exception Safe >Value Semantics >Raw pointers are just an accident that waits to happen

Use RAII classes such as Smart pointers proactive

Richard Peschke | Coding Guidelines | | Page 48 Clean interfaces through RAII

Without RAII processor* getProcessor(const char* N am e); processor* getReference2Processor(const char* N am e); > O w ned objects and optional references to objects have the sam e type

With RAII unique_ptr< processor> getProcessor(const char* N am e); processor* getReference2Processor(const char* N am e); > O bject ow nership is indicated by the type  type safe

Richard Peschke | Coding Guidelines | | Page 49 Clean interfaces through RAII

unique_ptr< processor> ow nedO bject processor ow nedO bject processor* optionalReference processor& nonO ptionalReference

Richard Peschke | Coding Guidelines | | Page 50 Functions:

What do you do when you write a function that is 1000 lines long? You are doing something wrong

Richard Peschke | Coding Guidelines | | Page 51 Code Reuse

void fun_A() { void fun_B() { if (pi= = 5) if (pi == 3) { { //1000 lines of very important code //1000 lines of very important code } } if (e == 3) if (e == 2) { { //////////////////////////////////////////////////// //////////////////////////////////////////////////// // starting doing stupid TASK A //////////////// // starting doing stupid TASK A //////////////// //////////////////////////////////////////////////// ////////////////////////////////////////////////////

//1000 lines of very important code //1000 lines of very important code

//////////////////////////////////////////////////// //////////////////////////////////////////////////// // finishing stupid TASK A //////////////// // finishing stupid TASK A //////////////// /////////////////////////////////////////////////// /////////////////////////////////////////////////// } } } }

Richard Peschke | Coding Guidelines | | Page 52 Code Reuse

void fun_A() >Make “Task A” a function >Call the function if(pi= = 5) Do …

if(e= = 3) Task A

void fun_B()

if(pi= = 3) Do …

if(e= = 2) Task A

Unique code Shared code Richard Peschke | Coding Guidelines | | Page 53 Code Reuse

void fun_c() { void fun_d() { std::vector< int> vec = getVector(); std::vector< int> vec = getVector(); for (auto& e: vec) for (auto& e : vec) { { /////////////////////////////////////////////////// //////////////////////////////////////////////////// // starting doing stupid TASK C /////////////// //== = = > starting doing stupid TASK D <= = = = ////////// /////////*********************************///////// /////////*********************************//////////

//1000 lines of very important code //1000 lines of very important code

/////////*********************************///////// /////////*********************************////////// // finishing stupid TASK C //////////////// //== = = > finishing stupid TASK D <= = = = /////////// /////////////////////////////////////////////////// //////////////////////////////////////////////////// } }

} }

Richard Peschke | Coding Guidelines | | Page 54 Code Reuse: Generic Programming

void fun_c() >Function pointer . Signal/Slot (QT/ROOT) for(auto& e:vec) Task C . Function hooks >Inheritance >Templates

void fun_d()

for(auto& e:vec) Task D

Unique code Shared code Richard Peschke | Coding Guidelines | | Page 55 Generic Programming: Inheritance

class base{ class Task_C :public base { public: public: void ProcessVector() { virtual void ProcessElem entOfVector(int) override final std::vector< int> vec = getVector(); { for (auto& e : vec){ // Task C ProcessElem entOfVector(e); } } }; } virtual void ProcessElem entOfVector(int) = 0; };

Base::ProcessVector()

Task_C::ProcessElem entOfVector(int)

Unique code Shared code Richard Peschke | Coding Guidelines | | Page 56 Lets Talk about the Pit

Richard Peschke | Coding Guidelines | | Page 57 Where is the Pit?

Your goals in live

This is the Pit

Richard Peschke | Coding Guidelines | | Page 58 One function has one task!

Every function should stay on one level of abstraction

D Done! OR: A B

D O A S B T O N

Richard Peschke | Coding Guidelines | | Page 59 What you wanted to do

H i Loop g Read- Find- h Display

Line Once l e v If(object is empty){ If (Keyword found){ e L

n Display(input)

o next Task i Stop t c a r t

s } else { Stop b next Task

A Read_from(Object) } else { next

w Continue o Task Continue L } Continue

} Richard Peschke | Coding Guidelines | | Page 60 What was written

> The Levels of Abstractions are totally mixed up Loop Welcome to your MC Escher Program While(!stopped){

} Richard Peschke | Coding Guidelines | | Page 61 What was written

> The Levels of Abstractions are totally mixed up Loop Enjoy your debugging! While(!stopped){

} Richard Peschke | Coding Guidelines | | Page 62 How could It look like

In Actual Code In Pseudo Code void readFile(const char* nam e, const char* keyW ord) { void readFile() { // initialize init(); // open file auto inFile = fileReader(nam e); loop{ if (!inFile.isValid()) readLine(); throw FileIsInvalid; CheckForKeyW ord(); D isplay(); // Loop } w hile (true) { Replace Loop Condition end(); //clean up if needed // begin file reader } if (inFile.EndO fFile()) break; auto line = inFile.getLine(); // end file reader

// Begin check for keyw ord if (KeyW ordFound(line, keyW ord)) continue; // end check for keyw ord

// begin Display D isplay(line); // end Display } //end } Richard Peschke | Coding Guidelines | | Page 63 Every Program in a Nutshell

>Responsibility Handlers > One can combine this them > One must not blend them! . Int / double / vector / > The inner structure is never visible for the smart_ptr / file_handler / outside world > Make sure that every part of your program is GUI_button / GUI_DrawPanel always recognizable as what it is >Data Container > There is no fourth type! >Day / Month / Year* / Processors: > finite state machine . Events / Hits / Tracks / . Begin / run / end Configurations > Processors can be chained together > >Tasks / Processors Processors can contain other processors >Load file / save file / . filter events / event listener

Event Event Event File reader File writer A A B Filter Converter File Handler MH MH MH File Handler MH MH MH

Richard Peschke | Coding Guidelines | | Page 64 * From the example at the beginning MH = memory handler Don’t write Hybrid Classes

What you expect: What you get:

Richard Peschke | Coding Guidelines | | Page 65 Error Handling

>Tasks can go wrong!

bool SendReceivePacket(...) { >Error handling is not about M utexLock m (...); Actual Code try { in EUDAQ writing try catch m .Lock(); } catch (...) { // sw allow it everywhere! } SendPacket(...); return ReceivePacket(...); >Error handling is about: } . Leaving the system in a defined state . Clean up your mess . Communicating errors . (user/ caller)

Richard Peschke | Coding Guidelines | | Page 66 Return Values VS Exceptions

Richard Peschke | Coding Guidelines | | Page 69 Error Safe Code with Exceptions

e d o c spot the d a difference b ll ti S

Richard Peschke | Coding Guidelines | | Page 70 RAII RAII

Richard Peschke | Coding Guidelines | | Page 73 C - Style Interfaces What could go wrong?

Richard Peschke | Coding Guidelines | | Page 74 What do I mean by C – Style Interfaces?

void func(int* vec, size_t lengthO fVec); >Pointer + length Interfaces >No visible connection void func1(size_t, int*, size_t, int*, size_t, int*, size_t, int*, between pointer and its size_t, int*, size_t, int*, size_t, int*, size_t, int*, size_t); length (besides the name)

>Can easily become unmaintainable >Hard to read >Very hard to reason about

Richard Peschke | Coding Guidelines | | Page 75 C - Style Interfaces

Neil Macintosh Richard Peschke | Coding Guidelines | | Page 76 Annotating Variables to show Intention

Ugly Ugly

Pretty awesome

Neil Macintosh Richard Peschke | Coding Guidelines | | Page 77 What is the problem?

>Pointer & length have no connection void func(size_t, int*); >Can we not just use a class? class array_view { public: void fun(array_view ); array_view (size_t, int*); int* data(); size_t size(); private: size_t m _size; int* m _data; };

Richard Peschke | Coding Guidelines | | Page 78 Use Objects Which handle the Pointer and Length

Neil Macintosh Richard Peschke | Coding Guidelines | | Page 79 The Caller Side

Obey: Document: >Single definition rule >On code level why you are doing . Avoid magic numbers . Avoid copy and paste something > Single responsibility rule >On library level how to use the . If it doesn't fit on one screen it does more then one thing. classes/functions Use: >Provide Examples > Modern code Think Data: . Less typing/easier to read > RAII (smart pointer/ vector) >How can you visualize your . Avoid RAW memory access. data? Avoid: . Control plots! >C-style … Anything >Separate between how you store data on disk and how you use it in code

Accessing elements of an array >What is the difference? >If you access elements in order the computer can guess what is the next element and load it before you access it >Your computer acts like it has 3 levels of cache of almost infinite size

Richard Peschke | Coding Guidelines | | Page 84 Herb Sutter Memory Memory Memory Memory Memory Memory Memory

Richard Peschke | Coding Guidelines | | Page 85 Herb Sutter How to understand code?

Richard Peschke | Coding Guidelines | | Page 86 How to understand code?

void Unknown_Function(int *a, int b) { int i, j, p, t; >What does this function if (b < 2) return; p = a[b / 2]; do? for (i = 0, j = n - 1;; i++, j--) { while (a[i] < p) >What are the input i++; while (p < a[j]) parameters? j--; if (i >= j) break; t = a[i]; a[i] = a[j]; a[j] = t; } Unknown_Function(a, i); Unknown_Function(a + i, b - i); }

Richard Peschke | Coding Guidelines | | Page 87 Think Data!

>What does the Function do?

 It’s a sort algorithm! >Is the sort stable?  Yes

Unknown Function

Richard Peschke | Coding Guidelines | | Page 88 Working Provide Examples

int main () { >What does this function do? int a[] = {4, 65, 2, -31, 0, 99, 2, 83, 782, 1}; int size_of_a = sizeof a / sizeof a[0]; >What are the parameters? Print(a, size_of_a); //a = 4, 65, 2, -31, 0, 99, 2, 83, 782, 1

Unknown_Function(a, size_of_a );

Print(a, size_of_a); // a = -31 0 1 2 2 4 65 83 99 782 return 0; }

Richard Peschke | Coding Guidelines | | Page 89 Template Example:

>Generic function >Works on all containers . If it doesn’t work on your favorite container, provide begin, end and getContent functions and it will work >No inheritance needed >Flexible

Richard Peschke | Coding Guidelines | | Page 90 Use of ‘const’ in Functions Return Values

char *Function1() You cast away the const-ness of your { return “Some text”;} return value

Function1()[1]=’a’; runtime error

Fix:

const char *Function1() compile time error { return "Some text";} Don’t cast away the const-ness of your return type

http://duramecho.com/ComputerInfor mation/WhyHowCppConst.html Richard Peschke | Coding Guidelines | | Page 91 Two types of classes

Data Structures Objects >Are just storage containers >Don’t show internal structure >Keep variables together >Have tasks >Show internal structure >Provide function to get/set inner >Don’t provide function to status manipulate content >Preserve invariance >Global functions can manipulate . Resource handling content >Objects derive functionality from  Easy to implement new functionality other objects  Easy to implement new Objects

Richard Peschke | Coding Guidelines | | Page 92 Two types of classes

Data Structures Objects

Richard Peschke | Coding Guidelines | | Page 93 Object Or Data Structure?

>A getter function doesn’t make an Object! >If you have a getter function like this you have direct access to the data!

Richard Peschke | SCT Upgrade | 20.11.2013 | Page 94 Objects and Invariance

>In the Process the system is in a >Multi thread safe status where “amount” is missing >Error handling in the system. >If another process checks the status at the same time it gets a wrong answer >No data recovery if an error happens >Hybrid Class

Richard Peschke | Coding Guidelines | | Page 95 CONST basic explanation

What does it mean? >YOU can only read from this object >Somebody else might be able to change the value

Where should you use it? >Everywhere as long as the compiler doesn’t complain

Richard Peschke | Coding Guidelines | | Page 96 what does const do with variables?

> const int * Constant2 . declares that Constant2 is a variable pointer to a constant integer and > int const * Constant2 . is an alternative syntax which does the same > int * const Constant3 . declares that Constant3 is constant pointer to a variable integer > int const * const Constant4 . declares that Constant4 is constant pointer to a constant integer.

http://duramecho.com/ComputerInfor mation/WhyHowCppConst.html Richard Peschke | Coding Guidelines | | Page 97 Const and Function parameters

Choose your favorite Function

i) t & ) t in i ns t & (co in d f2 3( voi f i) id t vo (in f1 id vo

Richard Peschke | Coding Guidelines | | Page 98 Const and Function parameters void f1(int i) void f2(const int &i) >you need a copy of “I” >You don’t need a copy “I” and and >You don’t want to change >You don’t want to change the input parameter the input parameter

void f3(int &i) >You want to change the input parameter

Richard Peschke | Coding Guidelines | | Page 99 How to program

T h is i s no t p ro gr am m i ng

Richard Peschke | Coding Guidelines | | Page 100 Types and why do we care

>Now he can meet his Wife in time and space >Nothing can go wrong now

Well never mind! But at least they met each other… Richard Peschke | Coding Guidelines | | Page 101