Skip to content

Latest commit

 

History

History
145 lines (117 loc) · 7.53 KB

TODO.md

File metadata and controls

145 lines (117 loc) · 7.53 KB

TO DO List for Panel-base Demos for DiffAnnotator

This TODO list is about files (and planned files) in notebooks/panel/ subdirectory (and maybe, incidentally, also in planned demo/panel/ directory).

All files discussed here, be it Jupyter Notebooks (*.ipynb) or Python scripts (*.py), can and should be run with Panel, e.g. via panel serve.

TO DO List for 00-panel-simple-Matplotlib.ipynb Jupyter Notebook

This Jupyter Notebook is meant to demo plotting with matplotlib in a Panel app.

TO DO List for 01-timeline.ipynb Jupyter Notebook

This Jupyter Notebook serves as a demo for analyzing the timeline, that is what can be extracted from the timeline.*.purpose-to-type.json.

General

  • switch to interactive Matplotlib panes with the help of ipympl (if it starts working correctly), or to some other interactive plots libraries
    • hvPlot (hvplot.pandas, maybe hvplot.xarray),
      using either Bokeh or Plotly backend
    • Bokeh
    • Plotly
    • Vega via [Altair][] (possibly with VegaFusion to provide serverside scaling, if there are more than 5000 rows of data)

Bugs and issues

  • resolve the issue with counts for diff.n_rem, diff.n_mod, and diff.n_add not matching counts for -:count and +:count
  • make the app / notebook more robust with respect to where inside the directory structure panel serve is started (i.e. what is its working directory)
  • fix the performance issue with re-resampling and thus re-computing -/+ [%] multiple times after single change of selected column with a Select widget
  • add / uncomment @pn.cache decorators to improve performance

New panes

  • possibly optional pane or set of pane to help with debugging:
    • Debugger widget, plus the use of logging standard library (or solution compatible with it, like structlog)
    • JSONEditor widget in mode="view" to show the original JSON data
    • Tabular widget (which is replacement for DataFrame widget) to show all generated dataframes (original data, resampled data);
      or Perspective pane (which is replacement for DataFrame pane)
  • use Tqdm or other indicators (like Placeholder pane, or LoadingSpinner indicator) to show the information about current progress, and notify about work in progress

New controls

  • select one of available timeline*.json file with the Select or FileSelector widget
  • upload custom timeline*.json file with FileInput and/or FileDropper
  • select one of repositories in timeline*.json, if there is more than one
  • select date range to plot and analyze
    • use DateRangeSlider widget coupled with DatetimeRangePicker to select date range;
      default value would be the date range for a given author, the full range for widget would be maximum date range for all authors available in the JSON file
    • use Select widget to select one of pre-defined ranges, like in GitHub Insights, for example for all (full date range), last month, last 3 months, last 6 months, last 12 months, last 24 months
    • use RadioButtonGroup or ToggleGroup with behavior="radio" to switch between years, with default to the last year (today up to 1 year ago), like in GitHub Developer Overview page (perhaps if "Select Year" is selected)
  • IntSlider widget to select figure size (or two sliders, one for height, one for width)
    or rather make it work the one that exists in the notebook (instead of leaving it disabled)
  • Select, RadioButtonGroup, ToggleGroup, or Switch to select between using 'author.date' or 'committer.date' as the index, and for resampling
  • Toggle, or Select, or Switch to drop commits that are merges, or to drop commits that are either merges or root commits (leaving only commits with exactly 1 parent)
  • switching between linear, logarithmic, symlog, logit, etc. scales:
    • Select, or Switch, Toggle, etc. to switch between linear and symlog scale for aggregated number of commits, aggregated number of lines, and for histograms thereof (scale of values i.e. bin sizes for histograms, i.e. x-axis)
    • Select, or Switch, Toggle, Checkbox, etc. to switch between linear and log/symlog scale for the counts of values in the histograms (i.e. y-axis)
    • Toggle/Switch, integrated into Card header, to switch between log (default) and linear scale for color (v-scale) for heatmap
    • maybe Toggle, Switch, ToggleGroup, or another value in Select, to use the logit scale for [%] of counts
  • colors, colormaps, and color gradients
    • Select for choosing between a few selected options for colormaps for heatmap plots, like Reds/Greens, which would include a choice for banded colormap (i.e. no smooth transition between values, the color is quantized)
    • Toggle, ToggleGroup, Checkbox, or Switch to select between solid fill and gradient fill (see e.g. StackOverflow: fill_between gradient)

New plots

  • plots for patch size and/or patch spread metrics
  • stacked area plot for -/+ line types, or for -/+ line types [%]
  • Sankey plot (flow plot) diagram between file purpose and line type, author and line type, or directory structure and line type, etc.
  • heatmap type of plot, with time with a week resolution on x-axis, and day of the week on the y-axis (or hour of day), similar to the heatmap plot in GitHub Developer Contributions
    • perhaps split into months, like in Assayo, or with end of months marked somehow in the year of activity
    • perhaps split into years, and laid out one below another, so that seasonal changes might be easier to find
    • Switch/Toggle to toggle between UTC dates, and localtime (with respect to day of the week)
  • histogram of timezones (should usually be 1-2 timezones, due to DST, unless the developer in question travels between timezones, or moves)
  • histogram of UTC or local timezone times of authorship / commit, or lines, or files, aggregated over specified hours (if using localtime, split between work hours, maybe-work hours, night hours, free time - including weekends)

Planned Jupyter Notebook: 03-compare.ipynb

This Jupyter Notebook is intended to compare different plots between two selected authors. Note that the data file with stats that is read from needs to include data for more than one developer.

See also main /TODO.md file.
See the "Related projects" section in the main /README.md file.