As a developer, I usually need to use Gecko profiler (or others profiling tools) to help out solving performance bugs, to run it while Raptor is running may reduce development works. Moreover, if the process is automatic, we're also apt to collect more profiling data than doing it manually. The proposal is to implement a simple script at early stage: the script only needs to start Raptor and profiler at the same time, which can be done with two shell commands. We also need to stop the profiler with captured data at the end of Raptor test, although this part I don't think it's easy to be implemented with pure shellscript. There are still lots of details to clarify, but I think If we can get this work well, with valuable profiled data and the workflow is simpler than previous, we may take a look to see how to push it in the future. The major difficulty is Gecko profiler needs some special flags for profiling bootstrapping performance, and to profile the device during a specific period with clear markers from Raptor. Other issues include how to set an option to turn this feature on/off dynamically (since in some cases we needn't profiling), and how to collect the data and even display it on remote server properly. But if people think this would help us to diagnose performance issues in the future, I think we should consider to turn it on for automation tests. For example, it can be suspend at rural integration tests, but if any performance regression is triggered, it should re-run the test with profiling enabled.
[Tracking Requested - why for this release]:
tracking-b2g: --- → backlog
Summary: Raptor/Profiler Integration → [Meta] Raptor/Profiler Integration
Status: NEW → RESOLVED
Last Resolved: 3 years ago
Resolution: --- → WONTFIX
You need to log in before you can comment on or make changes to this bug.