Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Visualizing Java code bases for JEEConf 2017

548 views

Published on

Are code lines added at high speed? Too many developers? How do you see the big picture? What processes are going on with your large code base? This presentation will show how to leverage existing code analysis tools (Cloc, Structure101, SonarQube) and combine them with data storage (ElasticSearch, Neo4j) and visualization tools (Gource, D3, Inkscape, Kibana) to at least make some sense out of millions of code lines and their history. During his consulting work, author often meets unfamiliar and at the same time large Java code bases that need quick analysis and input for decision making. That’s where visualizations come to into play by helping mining important knowledge directly from the code statistics.

Published in: Technology
  • Be the first to comment

  • Be the first to like this

Visualizing Java code bases for JEEConf 2017

  1. 1. 01
  2. 2. 02
  3. 3. Let's start! 03
  4. 4. Background Find new clients Face big code bases Need quick analysis Need quick results • • • • 04
  5. 5. Code size Google: 2 billions LOC Facebook: 61 million LOC Me: from 20K to 2M LOC • • • 05
  6. 6. I hate reading... 06
  7. 7. It's life, Jim, but not as we know it! 07
  8. 8. Code is data! 08
  9. 9. Well, big data 09
  10. 10. Snapshot 10
  11. 11. Temporal 11
  12. 12. Who uses SonarQube? 12
  13. 13. SonarQube Don't get me wrong: I love it! BUT... it's not that easy to get data in. It needs to be tuned for each team. It's not easy to make your team to use the data. Data is impersonal. • • • • • 13
  14. 14. First steps 14
  15. 15. Count the lines! 15
  16. 16. Size does matter Gives you an estimate (on how much reading is needed). Most of the code bases are polyglot. Ratio between languages can tell something. Ratio between test code, comments, blank lines is also interesting. • • • 16
  17. 17. cloc 17
  18. 18. Usage cloc ‐‐help cloc ‐‐write‐lang‐def=lang.defs 01. 02. 18
  19. 19. Usage cloc ‐‐csv       ‐‐quite       ‐‐progress‐rate=0      ‐‐ignored=files.ignored       ‐‐exclude‐dir=test,build      ‐‐read‐lang‐def=lang.defs       ‐‐out=data.csv      . 01. 02. 03. 04. 05. 06. 07. 08. 19
  20. 20. Language definitions Gradle     filter remove_matches ^s*//     filter remove_inline //.*$     filter call_regexp_common C     extension gradle     3rd_gen_scale 4.10 01. 02. 03. 04. 05. 06. 20
  21. 21. Where are the pictures? 21
  22. 22. We have stats, let's plot them!22
  23. 23. Excel? Probably not. 23
  24. 24. Pie charts are boring! 24
  25. 25. Infographics! 25
  26. 26. d3.js 26
  27. 27. d3.js Javascript library for data visualizations Tons of examples Many libraries built on top of d3.js Several books • • • • 27
  28. 28. Useful charts Calendar View ­ https://bl.ocks.org/mbostock/4063318 Buble Chart ­ https://bl.ocks.org/mbostock/4063269 Hierarchical Edge Bundling ­ https://bl.ocks.org/mbostock/1044242 • • • 28
  29. 29. Demo 29
  30. 30. Many alternatives vis.js raphael.js sigma.js many, many, many more • • • • 30
  31. 31. Tableau Public 31
  32. 32. Tableau Public 32
  33. 33. Tableau Public 33
  34. 34. Design your own! 34
  35. 35. SVG + Inkscape 35
  36. 36. Home­made Inforgraphics 36
  37. 37. Code City 37
  38. 38. Code City 38
  39. 39. Implementations eXcentia plugin for SonarQube JSCity • • 39
  40. 40. Temporal analysis 40
  41. 41. GitHub 41
  42. 42. GitHub 42
  43. 43. FishEye 43
  44. 44. Gource Software projects are displayed by Gource as an animated tree with the root directory of the project at its centre.  Directories appear as branches with files as leaves.  Developers can be seen working on the tree at the times they contributed to the project. “ 44
  45. 45. Launch gource gource ‐s 0.1 ‐1280x720  gource ‐‐output‐custom‐log log1.txt  gource ‐s 0.1 ‐1280x720 log1.txt 01. 02. 03. 45
  46. 46. Log format 1444125624|Luke Daley|M|/ratpack‐core/.../WiretapPublisher.java 1444125624|Luke Daley|M|/ratpack‐core/.../YieldingPublisher.java 1444306114|Stian Lindhom|M|/ratpack‐manual/.../13‐http.md 1444312172|Andrey Antukh|M|/ratpack‐core/.../WebSocketEngine.java 01. 02. 03. 04. 46
  47. 47. Demo 47
  48. 48. Code Maat Code Maat is a command line tool used to mine and analyze data from version­control systems “ 48
  49. 49. Launch Maat git log ‐‐pretty=format:'[%h] %aN %ad %s'      ‐‐date=short ‐‐numstat ‐‐after=YYYY‐MM‐DD 01. 02. 49
  50. 50. Launch Maat maat ‐l git.log ‐c git ‐a summary  maat ‐l ratpack_evo.log ‐c git ‐a revisions > ratpack_freqs.csv python scripts/merge_comp_freqs.py ratpack_freqs.csv ratpack_lines 01. 02. 03. 50
  51. 51. Analysis types abs­churn, age, author­churn, authors, communication, coupling, entity­churn, entity­effort, entity­ownership, fragmentation, identity, main­dev, main­dev­by­revs, messages, refactoring­main­dev, revisions, soc, summary 51
  52. 52. Demo 52
  53. 53. Let's talk big data 53
  54. 54. ElasticSearch Distributed, scalable, and highly available Real­time search and analytics capabilities Sophisticated RESTful API • • • 54
  55. 55. Index data $ curl ‐XPUT 'http://localhost:9200/gitlog/commit/123345' ‐d '{   "commitId" : "123345",   "timestamp" : "2009‐11‐15T14:12:12",   "message" : "git into es" }' 01. 02. 03. 04. 05. 55
  56. 56. Kibana Flexible analytics and visualization platform Real­time summary and charting of streaming data Intuitive interface for a variety of users Instant sharing and embedding of dashboards • • • • 56
  57. 57. Kibana 57
  58. 58. Demo 58
  59. 59. Structure 59
  60. 60. Structure101 60
  61. 61. Demo 61
  62. 62. Graphs are everywhere! 62
  63. 63. jQAssistant jQAssistant is a QA tool which allows the definition and validation of project specific rules on a structural level.  It is built upon the graph database Neo4j and can easily be plugged into the build process to automate detection of constraint violations and generate reports about user defined concepts and metrics. “ 63
  64. 64. jQAssistant jqassistant scan ‐f binaries jqassistant server 01. 02. 64
  65. 65. Query your code MATCH   (class:Class)‐[:DECLARES]‐>(method:Method) RETURN   class.fqn, count(method) as Methods ORDER BY   Methods DESC LIMIT 20 01. 02. 03. 04. 05. 06. 07. 65
  66. 66. Query different versions match   (artifact:Artifact)‐[:CONTAINS]‐>(type:Type) return   artifact.fileName as Artifact, collect(type.fqn) as Types 01. 02. 03. 04. 66
  67. 67. neo4j 67
  68. 68. Conclusion 68
  69. 69. Adam Tornhill 69
  70. 70. Hotspot Analysis use hotspots to identify maintenance problems. use hotspots for risk management. hotspots point to code review candidates. hotspots are input to exploratory tests. • • • • 70
  71. 71. Conclusion Extract data from your code! Visualize it and search for hot spots! Search for new facts and knowledge! Become data scientist or data journalist! • • • • 71
  72. 72. Next time... 72
  73. 73. when you... 73
  74. 74. push your code 74
  75. 75. REMEMBER 75
  76. 76. 76
  77. 77. Reading material 77
  78. 78. Your Code as a Crime Scene 78
  79. 79. Metrics and Models in SQE 79
  80. 80. Software Metrics and Metrology 80
  81. 81. Links: images http://abstrusegoose.com/432 http://camarenaphoto.tumblr.com/post/112238079516/its­life­jim­but­ not­as­we­know­it­spock http://technology.ie/big­data­looks­like/ http://www.informationisbeautiful.net/visualizations/million­lines­of­ code/ http://githut.info/ http://emmanueloga.com/2013/10/07/Graphs­are­Everywhere­An­ overview­of­GraphConnect­San­Francisco­2013.html • • • • • • 81
  82. 82. Links: tools https://github.com/AlDanial/cloc http://gource.io/ http://wettel.github.io/codecity­wof.html https://github.com/adamtornhill/code­maat http://d3js.org/ http://visjs.org/ • • • • • • 82
  83. 83. Links: tools http://www.sonarqube.org/ https://www.atlassian.com/pt/software/fisheye/overview https://www.elastic.co/products/elasticsearch https://www.elastic.co/products/kibana http://jqassistant.org/ http://neo4j.com/ • • • • • • 83
  84. 84. Links: tools https://github.com/ThoughtWorksStudios/saikuro_treemap https://www.youtube.com/watch?v=iiIytERhV9o https://codescene.io http://www.adamtornhill.com/articles/software­ revolution/part1/index.html • • • • 84
  85. 85. That's all! 85
  86. 86. Thank you! 86
  87. 87. 87

×