Does Your Code Measure Up?

Transcription

1 Does Your Code Measure Up? By: Adam Culp

2 2 About me PHP 5.3 Certified Consultant at Zend Technologies Organizer SoFloPHP (South Florida) Organized SunshinePHP (Miami) Long distance (ultra) runner Judo Black Belt Instructor

3 Fan of iteration 3 Pretty much everything requires iteration to do well: Long distance running Judo Development Evading project managers Quality!

4 How To Measure? 4 Not productive

5 How To Measure? 5 More accurate, quantifiable

6 Why Measure? 6 Highlight bugs

7 7 Why Measure? Highlight bugs Improve quality Easier onboarding Less reading, more writing Testable

8 Why Measure? Highlight bugs Improve quality 8 Easier onboarding Less reading, more writing Testable Satisfied customers Faster development Less broken

9 Why Measure? Highlight bugs Improve quality 9 Easier onboarding Less reading, more writing Testable Satisfied customers Faster development Less broken Personal pride

10 Why Measure? Highlight bugs Improve quality 10 Easier onboarding Less reading, more writing Testable Satisfied customers Faster development Less broken Personal pride Higher salary

11 What To Measure? Cyclomatic complexity 11 The count of the number of linearly independent paths through the source code. - wikipedia Decision points Less than 10 (personally less than 6)

12 12 Cyclomatic Complexity Example

13 What To Measure? Cyclomatic complexity Duplicate code 13 Rule of 3

14 14 What To Measure? Cyclomatic complexity Duplicate code Long classes Less than 1,000 lines Classes solve a (1) problem

15 What To Measure? Cyclomatic complexity Duplicate code Long classes Class Complexity 15 Less than 50

16 16 What To Measure? Cyclomatic complexity Duplicate code Long classes Class complexity Long methods Less than 100 (personally less than 20) Method should do one thing

17 17 What To Measure? Cyclomatic complexity Duplicate code Long classes Class complexity Long methods Unused variables

18 What To Measure? Cyclomatic complexity Duplicate code Long classes Class complexity Long methods Unused variables Lack or overuse of comments 18 Clear, concise, not explain bad code

19 19 Comment Example

20 20 What To Measure? Cyclomatic complexity Duplicate code Long classes Class complexity Long methods Unused variables Lack or overuse of comments Heavy global usage

21 21 What To Measure? Cyclomatic complexity Duplicate code Long classes Class complexity Long methods Unused variables Lack or overuse of comments Heavy global usage Npath complexity Possible paths through code Less than 200 paths

22 What To Measure? Cyclomatic complexity Duplicate code Long classes Class complexity Long methods Unused variables Lack or overuse of comments Heavy global usage Npath complexity Much, much, more 22 Code smells

23 Code smells 23 What are smells? Indications of spoiled code nearby Not conclusive The smell is not bad

24 Code smells 24 Smells hinting a refactor may be needed: Duplicate Code (rule of 3) Long Methods Large Class Long Parameter (argument) List Divergent Change cascade change to accommodate another Shotgun Surgery change ripples as bugs Feature Envy method uses parts from other class Switch Statements sacrifice polymorphism

25 Code smells 25 Cont'd: Lazy Class class not doing much Speculative Generality something built for possible future Temporary Field/Variable Message Chains object asking object asking object Middle Man directors in place but serve no real purpose Inappropriate Intimacy classes share private parts Data Class getters and setters, but nothing else Comments where comments cover bad code

26 Tools 26 PHPqatools.org

27 Tools PHPqatools.org 27 PHPLoc

28 28 PHPLoc Results

29 Tools 29 PHPqatools.org PHPLoc PHP_Depend

30 PHP_Depend Result Graphs and XML output ANDC AHH NOC NOM LOC NOP CYCLO Low Average High NOM CALLS FANOUT Generated by PHP_Depend

31 Tools 31 PHPqatools.org PHPLoc PHP_Depend PHP Copy/Paste Detector

32 32 PHP Copy/Paste Detector Result

33 Tools PHPqatools.org PHPLoc PHP_Depend PHP Copy/Paste Detector PHP Mess Detector 33 Codesize

34 34 PHP Mess Detector Result (codesize)

35 Tools PHPqatools.org PHPLoc PHP_Depend PHP Copy/Paste Detector PHP Mess Detector 35 Codesize Unused

36 Tools PHPqatools.org PHPLoc PHP_Depend PHP Copy/Paste Detector PHP Mess Detector Codesize Unused PHP Dead Code Detector 36 Searches for code no longer used

37 Tools PHPqatools.org PHPMetrics 37

38 38 PHPMetrics Results

39 39 Tools PHPqatools.org PHPMetrics PHP_Codesniffer Create rules/sniffs to ensure standards are followed From CLI, IDE, or via SCM hooks

40 Tools PHPqatools.org PHPMetrics PHP_Codesniffer Code Climate Build from Git/Github repo Open = free Private = $ GPA like in high school, simple 40

41 Code Climate Build from Git/Github repo 41 Open = free Private = $

42 Tools PHPqatools.org PHPMetrics PHP_Codesniffer Code Climate Zend Server and Z-Ray 42 Debugging, Code tracing, Error reporting

43 Realtime profiling 43 Z-Ray

44 Tools PHPqatools.org PHPMetrics PHP_Codesniffer Code Climate Zend Server and Z-Ray Apache Bench (apache2-utils in apt) 44 How much traffic can you handle?

45 45 Apache Bench Result

46 46 Conclusion Measure all the things! Don't fear results, share them Reduce complexity Leave code cleaner than you got it Learn to smell problems Use refactoring to fix shortcomings Love iteration!

47 Thank you! Code: Please rate at: Adam Culp