grapher kwargs modification

[fdion]

When I implemented symbols for lines, I relied on what was already there, namely "data_frame" in kwargs. This was defined in on_dragger_finished as kwargs = {"data_frame": df}in grapher.py.

In the current develop branch, I see kwargs is now kwargs = {} in on_dragger_finished. Nothing else sets it.

I'm not sure if that was needed for some other part of the code, but this break:

• line plots
• bar plots

Maybe that was an artifact of a merge? Trying to figure out wheter that is the bug, or if we need to fix line and bar in a different way, or is there something else missing somewhere else.

Figured I'd start a discussion instead of a bug, cause I don't know exactly how to formulate the bug report. Also not sure what section I should put this in, so I'm putting it in Q&A

[adamerose]

Looks like I broke this in one of my commits adding code history support to the Grapher. I needed to be able to call .add_history_item inside the graphing functions to record the equivalent Python snippet for each one, this meant passing a PandasGuiDataFrameStore object to the plotting functions instead of a simple DataFrame.

To work with this new structure, I access the DataFrame inside the plot functions using pgdf.df instead of kwargs['data_frame'], just forgot to update the logic you added to line and bar. I think I just fixed it now though if you want to double check.

[fdion]

I figured something needed more flexibility and/or multiple calls.

In regards to line, this works fine, but for the bar plot, key_cols is not what you'd think and it is throwing a KeyError. Using one of your demo dataset, selecting Speed for x you get not Speed in key_cols, but ('S', 'p', 'e', 'e', 'd'). Are there hierarchical kwargs passed to bar?

[adamerose]

Yeah if you have multiple columns dropped in a category like x then the value in kwargs will be a list of strings instead of a string. The logic at the top of line accounts for this and I tried simplifying it to a one-liner in bar but apparently didn't work.