Debugging 2013- Lars pedersen
0 likes218 views
This document discusses how to spend less time debugging software by implementing quality practices up front. It recommends conducting peer code reviews, where another developer reviews code changes and provides feedback before integration. Code reviews are most effective when the author pre-reviews their own code and reviewers spend 30-60 minutes on each review. The document also advocates for establishing continuous integration processes using tools like Git, Gerrit and Jenkins to automatically build and test all code commits, making bugs easier to detect and fix early. Implementing these practices can help catch up to 90% of bugs during development rather than later on, saving significant costs.
![Because
Average defect detection rate is 25 percent for
unit testing, 35 percent for function testing, and
45 percent for integration testing. [McConnell93]
Furthermore, the average effectiveness of
design and code inspections are 55 and 60
percent. [McConnell93]](https://image.slidesharecdn.com/lars-pedersen-131029022601-phpapp02/85/Debugging-2013-Lars-pedersen-6-320.jpg)
![Peer Code Review
● Beneficial for “quality of review” if the author
pre-reviews and leaves comments for
subsequent reviewers [Cohen2006]
● The longer a reviewer spends on the initial
read-through, the more defects will ultimately
be found [Uwano2006]
● Defect detection drops dramatically after ~60
minutes, to zero after ~90 minutes [Dunsmore2000]](https://image.slidesharecdn.com/lars-pedersen-131029022601-phpapp02/85/Debugging-2013-Lars-pedersen-8-320.jpg)
![Peer Code Review
● Long methods and loops are very time
consuming to understand [Uwano2006]
● Review 100 to 300 LOC at a time, in 30-60
minutes chunks, with a break between each
sitting [Cohen2006]
● Spend at least 5 minutes reviewing a single
line of code [Cohen2006]
● Limit reviewing to 1 hour per day [Ganssle2009]](https://image.slidesharecdn.com/lars-pedersen-131029022601-phpapp02/85/Debugging-2013-Lars-pedersen-9-320.jpg)
![Our experiences - Low Gain Example
[Anonymized and obscured]](https://image.slidesharecdn.com/lars-pedersen-131029022601-phpapp02/85/Debugging-2013-Lars-pedersen-19-320.jpg)
![Our experiences - High Gain Example
[Anonymized and obscured]](https://image.slidesharecdn.com/lars-pedersen-131029022601-phpapp02/85/Debugging-2013-Lars-pedersen-20-320.jpg)
More Related Content
What's hot (20)
Viewers also liked (16)
Similar to Debugging 2013- Lars pedersen (20)
Recently uploaded (20)
Debugging 2013- Lars pedersen
- 1. Quality Up Front How to spend less time debugging Lars Jarnbo Pedersen Co-founder Switch-Gears ApS
- 2. About me Partner at Switch-Gears ApS ● Transitioning companies from old development tools to modern Open Source ones Been working for Nokia 2001-2012 ● Release manager for S40 software for many years ● Part of a small team that introduced Git, Gerrit and Jenkins to S40 development resulting in BIG productivity gains
- 3. Pain in SW Development 50% COST OF BUGS of developer time is spent finding and fixing bugs* *http://www.businessweekly.co.uk/hi-tech/14898-software-bugs-cost-more-than-double-eurozone-bailout
- 4. Why? Programming is hard… ● you have to map the problem domain to the programming language in use ● its usually a collaborative effort ● time pressure ● best practises are not shared Bugs are discovered late… ● context is lost ● often it is not the programmer who introduced the bug, who is fixing it
- 5. Luckily 90% COST OF BUGS of the bugs can be caught during development* *http://www.codinghorror.com/blog/2006/01/code-reviews-just-do-it.html
- 6. Because Average defect detection rate is 25 percent for unit testing, 35 percent for function testing, and 45 percent for integration testing. [McConnell93] Furthermore, the average effectiveness of design and code inspections are 55 and 60 percent. [McConnell93]
- 7. How?
- 8. Peer Code Review ● Beneficial for “quality of review” if the author pre-reviews and leaves comments for subsequent reviewers [Cohen2006] ● The longer a reviewer spends on the initial read-through, the more defects will ultimately be found [Uwano2006] ● Defect detection drops dramatically after ~60 minutes, to zero after ~90 minutes [Dunsmore2000]
- 9. Peer Code Review ● Long methods and loops are very time consuming to understand [Uwano2006] ● Review 100 to 300 LOC at a time, in 30-60 minutes chunks, with a break between each sitting [Cohen2006] ● Spend at least 5 minutes reviewing a single line of code [Cohen2006] ● Limit reviewing to 1 hour per day [Ganssle2009]
- 10. Mental Model Platform Knowledge + Development Process = Platform Quality
- 11. Increase Platform Knowledge Write Readable Code!!! ● You spend much more time reading code than writing it - so make it nice to read! ● Refactor if necessary ● Include test-code Commit changes in logical units ● Always commit working code ● Never mix error-fixes and feature implementations Write descriptive commit messages ● Make sure the reason for the change is clearly explained
- 12. Optimize Development Process Make it simple on the developers, ● allow multiple workflows at the developers desk ● don’t let the release/integration process stop the development flow ● minimize manual process steps while keeping a strict process ● automatically build and test all commits ● require code-review of all commits
- 13. This is NOT rocket science! So why don’t we do it already? ● Waterfall development ● Code Reviews are perceived expensive ● Last generation of tools do not support this easily What is enabling it now? ● Agile development ● New breed of Open Source tools
- 14. Git / Gerrit / Jenkins
- 15. Git (DVCS) changes everything! Git is NOT subversion++ ● all operations except for push/pull are local => well suited for continuous integration ● allows for multiple workflows ○ development ○ integration / releasing ● is a first class citizen during development AND debugging! ○ gitk for code analysis ○ git bisect for debugging
- 16. Gerrit A developer codes - and asks for feedback
- 17. Peer Code Review Another developer reviews and comments
- 18. Continuous Integration Actually you saw it in action already! Good tools help you without getting in the way
- 19. Our experiences - Low Gain Example [Anonymized and obscured]
- 20. Our experiences - High Gain Example [Anonymized and obscured]
- 21. References Books ● ● Code Complete: A Practical Handbook of Software Construction, Second Edition by Steve McConnell (Jun 16, 2004) Clean Code: A Handbook of Agile Software Craftsmanship by Robert C. Martin (Aug 11, 2008) Articles ● ● ● ● ● SCM Ranking (Switch-Gears) CI Ranking (Switch-Gears) Best Kept Secrets of Peer Code Review (Smartbear) The Cost of Software Bugs (Business Weekly) Code Review - Just Do It (Coding Horror) Videos ● ● Linus Torvalds on Git (Google TechTalk) Development at the Speed and Scale of Google (InfoQ)