ZSI / Debugging

Write code without errors!

The End

But mistakes happen

A Bug

Debugging

Prevention

• Testing
• Code quality

There are two ways to write code:
write code so simple there are obviously no bugs in it,
or write code so complex that there are no obvious bugs in it.

— Tony Hoare

Collect information

Error Tracking

• issue/bug tracking
• automatic error logging
  • tools such as Sentry

Logging

• Logging system events
• Program execution tracing
• Where?
  • súbor, stdout, stderr, syslog
• Levels
  • debug, info, warning, error, critical

Tools

• Libraries
  • syslog(3)
  • python logging
• Operating system
  • syslog
  • systemd-journald
  • docker

What about growing log files?

Log Rotation

• Moving to another file
• Deleting old ones

What if a bug occurs anyway?

Don’t Panic

This is true for galactic hitchhikers and for developers.

– The Pragmatic Programmer

What do we need?

• Understand the cause
• Fix it
• Don't break anything
• Improve code quality
• Make sure the bug would not return

Proces

1. Reproduce
2. Diagnose
3. Fix
4. Reflect

1. Reproduce

Generally, if you can’t reproduce it, it’s almost impossible to fix.

— Anonymous programmer, Practices of Software Maintenance, Janice Singer

Reproduce

• Program
• Environment
• Input

Simplify Reproduction

• minimal
• deterministic
• automatic

Failing Test Before Fixing Code

Create a focussed test that reveals the bug before you try fixing it.

– The Pragmatic Programmer

2. Diagnose

Read the Damn Error Message

Most exceptions tell both what failed and where it failed. If you’re lucky you might even get parameter values.

– The Pragmatic Programmer

Scientific Method

• Hypotheses and experiments
• “One change at a time”

Instrumentation

• Adding code to monitor system behavior

Heisenbug

Divide and Conquer

• Binary search

Version Comparison

• Regressions
• git bisect

Focus on differences

Debugger

• Stepping
• Monitoring program status
• Verifying assumptions

Debugger

• Useful tool for understanding code
• Not for repeated experiments
  • lacks automation
• Less useful when using TDD

3. Fix

Fix

• Solve the problem
  • source of the problem and not the symptoms
• Do not cause regression
  • tests
• Maintain or improve quality
  • refactoring

4. Reflect

The six stages of debugging

1. That can’t happen.
2. That doesn’t happen on my machine.
3. That shouldn’t happen.
4. Why is that happening?
5. Oh, I see.
6. How did that ever work?
7. It’ll never happen again!

Reflection

• How could this happen?
• Don't let it happen again
  • automatic checks
  • refactoring
  • process change

Conclusion

• prevention
• Systematic approach
• Find the source of the problem
• Never let it happen again