Home

Note

The following is related to early thoughts on package development. Please see the official package documentation for support.

---

I'm guessing this will just be a place to put ideas and discuss things ~ Ben

On package potential

In the notes/proof_of_concept.py document, I provide some an example using svg. I have code for png as well, but currently, I think they have similar problems, and svg is better.

In either approach, a problem arises that you either need to rescale the images after they're created or have "small-than-expected" images situated in each block. For the later, examples of this can be examined by changing the proposed_scaling_both function to proposed_scaling and then changing the hjust and vjust parameters to examine different alignments.

When I was working on the png approach (though this applied to svg as well?) I imagined we could try to "recorrect" the text and object sizes (points, lines, etc) of the plots after knowing the scaling - though this is harder if the height and width scales are different.

Package design

I imagined this package trying have similar attributes to gridExtra, cowplot and patchwork.

General structure of each:

1. patchwork
   • allows for different "+", "|", "/" grammar to be used to combine graphics (https://patchwork.data-imaginist.com/articles/guides/assembly.html)
   • allows for multiple depths of plot structure
   • provides additive functions to define layouts
   • allows for themes and objects to be generally changed
2. gridExtra
   • allows for defining of layout_matrix (like in the notes/proof_of_concept.py's top, e.g.

   np.array([[1,1,2,2],
             [1,1,0,0]])

   • allows for title texts at top, bottom, left, right
   • allows for varying width of rows and columns
3. cowpatch
   • allows for subplot titles / labels
   • allows for different ways to define layouts than gridExtra (with text, e.g.

   """
   1122
   11##
   """