#Project Silk Profiler

To profile project silk, we need a very light-weight tool to analysis output data of gecko::silk and present those data in a visual way.

What's project silk?

Where is source code of silk?

#Prerequisite

• Install matplot
• Python 2.7
• html 1.16
• (Optional) Install iPython

Ubuntu setup
$ sudo apt-get install python-pip   # in case you have not ever installed any python packages
$ sudo pip install ipython          # install ipython
$ sudo pip install pyzmq
$ sudo pip install Pygments
Mac setup... Oops...I have forgotten

Although iPython is not a must have, I strongly recommend you install this advanced shell, which has qtconsole and better integration with matplot.

Steps

Here is an example of how to parse a log file($3) by a pattern file($2)

$ cd /path/to/silkprofilerloader.py
$ ipython qtconsole --pylab=inline

#In  qtconsole
In [1]: %rub silkprofilerloader.py ./sample/testpattern_pass.pattern ./sample/log.txt
In [2]: SP.profiler.Draw(SP.Histogram.Line)    # Draw line chart
In [3]: SP.profiler.Draw(SP.Histogram.Bar)     # Draw bar chart
In [4]: SP.profiler.Draw(SP.Histogram.All)     # Draw all supported chart
In [5]: SP.profiler.DumpSamples()              # Dump silk raw samples
In [6]: SP.profiler.Statistic(True)            # Dump statistic data

####Result SilkProfiler.Draw(SP.Histogram.Line) SilkProfiler.Draw(SP.Histogram.Bar)

SilkProfiler.DumpSamples()

1.  frame number = 196826 / touch event distance = 0.016
2.  frame number = 196827 / touch event distance = 0.030
3.  frame number = 196828 / touch event distance = 0.037
4.  frame number = 196829 / touch event distance = 0.029
5.  frame number = 196830 / touch event distance = 0.037
6.  frame number = 196831 / touch event distance = 0.024

SilkProfiler.Statistic()

Total samples      = 33
Mean value         = 0.0204848484848
Standard deviation = 0.0145458964579

#Testing ####doctest

$ cd /path/to/pysilk
$ python silkprofiler.py -v
$ python silkreportgenerator.py -v

####unittest To run test case, simply execute testsilkprofiler.py

$ cd /path/to/pysilk
# Run SilkProfiler testing suite
$ python testsilkprofiler.py
# Run SilkReportGenerator testing suite
$ python testsilkgenerator.py
....
----------------------------------------------------------------------
Ran 4 tests in 0.002s

OK

#TBD

• adb controller
• Integrate with Jerry's pref setting tool
• start/ stop logcat
• log splitter
• split a log file into different ones depend on setup principle
• More statistic result
• Save to DHTML