Howto talk:Troubleshooting Memory Issues
Jump to navigation
Jump to search
Some thoughts
- users need to be really specific about their use-case and bug reporting
- users need to come up with a narrowed-down test-case ideally (changing/excluding factors like location/scenery and aircraft)
- users need to come up with an easily reproducible test case exhibiting the problem
- pause vs. unpause effect (some systems being disabled/suspended when paused)
- effects of accelerated sim-time vs. time warp to speed up troubleshooting
- the test case should be based on fgfs standards, i.e. fgfsrc, preferences.xml, startup arguments
- complex flights should either be covered via pre-recorded flights and/or fgtape
- alternatively, scripted flights would work, too
- if the use-case depends on complex procedures, use an aircraft with Aircraft Checklists to help execute those in a reprodudcible fashion
- inevitably, given the lack of dedicated RAM tracking means in FG, this will be highly Linux specific for now - so need to look for windows based contributors wanting to help with extending the article accordingly
- discuss git bisect and especially scripting "git bisect run" via fgfs
- introduce ASAN
- explain C++ classes, ctors and dtors for tracking allocations via new (see my posting), use of SG_LOG for console output