Web Peformance

Repository to extract web page timeline information using chrome dev tools api

Steps

Desktop (Mac/Linux)

Generate trace:

python run_timeline_trace.py <path to list of urls> <path to output directory> <device type (mac|android)>

Analyse trace: (Group computation into separate categories)

python run_log_trace_metrics.py <path to top level directory containing all traces> <path to output directory>

Generate plot from parsed traces:

python generate_plot.py <path to parsed trace top level directory>

Android

• Connect android to the desktop system running these scripts
• Enabel USB Debugging
• Start adb server: adb devices
• Follow the same steps as above, with android as the the device type

Extract Trace numbers

• The instrumented code contains an event handler which automatically computes the number of functions executed and invoked at page load time.
• Once the instrumented page is loaded inside a browser:

node replay_script.js <path-to-store-metadata> <path-to-store-invocations>

• Repeat the same steps either for a different recording of the same web page, or for the same page with a different replay instance

• With the two invocation files generates using the two runs, generate the stats about the number of intersection function signatures:

node compareWebPages.js <path-to-first-invocation> <path-to-second-invocation>