UCSC Web Development Talk

Transcript

1. Shipping Software in the Real World Paulo Franca, President and CTO Mat Malone, Director of Engineering
2. testing testing testing testing @ monitoring monitoring monitoring monitoring @ shipping shipping shipping shipping @
3. Who is DataCare?
4. TESTING IS ABOUT TRUST: customers trusting you. you trusting your code.
5. ACCEPTANCE TESTING Solve the customer’s problem! AUTOMATED TESTING Does my code do what I think it does?
6. ACCEPTANCE TESTING @ DataCare HOW WE DO THINGS.
7. ACCEPTANCE TESTING @ DataCare HUMANS
8. ACCEPTANCE TESTING @ DataCare HUMANS DOMAIN EXPERTS
9. ACCEPTANCE TESTING @ DataCare HUMANS DOMAIN EXPERTS USER CENTRIC
10. ACCEPTANCE TESTING @ DataCare HUMANS DOMAIN EXPERTS USER CENTRIC INDEPENDENT
11. ACCEPTANCE TESTING @ DataCare HUMANS DOMAIN EXPERTS USER CENTRIC INDEPENDENT COMPLIANCE
12. WHAT’S WRONG WITH THIS?
13. I WOULD RATHER BE SHIPPING.
14. QUALITY IS NOT MYJOB.
15. automated testing work in progress
16. Software development is hard. Automated testing makes it easier.
17. every organization must ask themselves: make somebody else write tests OR do it yourself?
18. TEST DRIVEN DEVELOPMENT
19. write test first watch test fail code see test pass refactorcode
21. WHY TDD IS AWESOME: REASONS
22. WHY TDD IS AWESOME: CONFIDENCE Refactor as much as you want. Avoid regressions during development.
23. WHY TDD IS AWESOME: SELF-DOCUMENTING Tests describe what the code does. If the code passes, you know the spec is true.
24. WHY TDD IS AWESOME: DESIGN Forces you to think about your design before you code.
25. WHY TDD IS AWESOME: EFFICIENT Catch bugs early. Before acceptance testing. Before a customer complains.
26. OMG!! TDD IS SO RAD what could possibly be wrong?
27. WE ARE AUTOMATED TESTING IS THE FIRST CORNER YOU CUT.
28. RUN TIME If the tests take 2 days to run no one will use them. DISCIPLINE If test coverage is bad you won’t trust your tests results.
29. PHPUnit JUnit Test::Unit RSpec Selenium C++Unit
30. TESTING RECAP ACCEPTANCE TESTING SOLVE THE CUSTOMER’S PROBLEM AUTOMATED TESTING NO BUG LEFT BEHIND, EVER
31. TESTING RECAP questions on testing?
32. monitoring andtelemetry
33. IS YOUR APPLICATION UP? RIGHT NOW? HOW ABOUT NOW? HOW DO YOU KNOW?
34. How DataCareMonitors Ahshay
35. How DataCareMonitors Ahshay HARDWARE CPU, HARD DRIVES/RAID, POWER SUPPLY
36. How DataCareMonitors Ahshay HARDWARE CPU, HARD DRIVES/RAID, POWER SUPPLY SERVICES APACHE, MYSQL, NFS, POSTFIX
37. the is what really matters login click • click •click search •download •upload
38. Your automated test suite comes in handy here.
39. No one likes waking up at 3am for a false-positive.
40. No one likes waking up at 3am for a flapping service.
41. No one likes waking up at 3am for an unreliable host.
42. StillAlive Nagios CloudWatch Cacti Pager Duty Zenoss
43. MONITORING RECAP ACCURACY: DON’T BUG ME FOR NO REASON THE APP: MONITOR WHAT REALLY MATTERS
44. MONITORING RECAP questions on monitoring?
45. shipping web based software in the real world
46. BUILDING ENTERPRISE SOFTWAREMEANS… BUSINESS ANALYSIS ONE SIZE DOES NOT FIT ALL CUSTOMIZATION vs.CONFIGURATION
47. Customers need the same features, but all want them a little different: Customer A calls it a “case” Customer B calls it a “claim” Customer C calls it a “worker” Standard software is more profitablebut more difficultto sell.
48. business vs. academia LIKE BEFORE BUT DIFFERENT DON’T SPECIALIZE YOUR CODE TOO MUCH
49. business vs. academia DATABASE INTENSIVE PROGRAMMING
50. business vs. academia DID YOU KNOW RECURSION IS USEFUL?
51. business vs. academia data structures UNSUAL DATA STRUCTURES
52. BIG RECAP test •monitor •reality check
53. thank you! Slides available at: http://speakerdeck.com/u/matmalone/p/ucsc-web-development-talk