Big Ant Memory and Crash Diagnostics
Big Ant Memory and Crash DiagnosticsProject cover · screenshots to come
Internal tools for memory-allocation analysis and crash capture, packaging and investigation.
My role
Created the crash-reporting system as sole programmer; improved an established memory-analysis tool.
What it solves
Investigating memory allocations and recurring crashes required assembling evidence from dumps and affected runs.
What it delivers
- Programmers can inspect engine allocations in a focused analysis tool.
- Crash dumps and logs are collected for investigation.
- Callstacks and run logs accompany crash reports.
Key decisions
- Improved an existing CSV memory-analysis tool and allocation-map responsiveness.
- Captured crash dumps, logs and run evidence through Windows Error Reporting.
- Kept a local crash bundle when the network destination was unavailable.
Explore the technical details
Capabilities
- CSV-driven analysis of engine memory allocations
- Windows Error Reporting crash capture for dumps and logs
- Crash grouping to help identify recurring signatures
- Crash bundles with callstacks and run logs for investigation
Further decisions
- Moved symbol and callstack lookup off the UI thread.
- Grouped crash reports to help triage recurring signatures.
I improved an existing memory-analysis tool and created a separate crash-reporting system. Together, they help programmers inspect allocation data and collect crash evidence for investigation.
Related work: developer automation and QA prototyping.