Skip to content

Analysis refactorizing

Created by: valschneider

Here's a notebook containing every analysis dataframe/plot that are working: https://gist.github.com/valschneider/fe7ee87e4ff3a6e3e443a316d21c04c0

Changelog:

  • Instored a colourblind-friendly colour cycle, so that we don't have to think about which colours to use
  • Added a requires_events decorator that both asserts trace events are available and add them to the method documentation.
  • Factorized plotting code (creation/saving to file)
  • Documented the shape of the returned DataFrame for all df_* methods.
  • Made all plotting methods return the matplotlib.axes.Axes they created/manipulated
  • Added a load_tracking analysis module for... load-tracking related analysis
  • Un-camelCased some methods
  • Incinerated the other analysis modules and factorized their ashes.

Things left to do:

  • Assert that some analysis methods haven't been broken (notably check that master.df_latency == next.df_task_states)
  • Status analysis
  • Thermal analysis
  • Update uses of the analysis API in lisa/tests/kernel
  • Update uses of the analysis API in lisa/tests/lisa
  • Un-camelCase the remaining offenders
  • (probably in a later PR) Make sphinx generate an example plot for plotting methods' documentation.

Merge request reports