WIP: basic outline of base version of repel

[MichaelChirico]

Some initial work on #156

The skeleton is there -- API I envision is expose repel_text (or maybe text_repel) as the "repel" version of graphics::text.default

TODO:

• [x] Implement pos argument
• [x] Implement offset argument
• [ ] Thorough testing of expected outputs
• [ ] Don't draw segments if the text "hasn't moved" (?)
• [ ] Examples
• [ ] NEWS
• [ ] Clean-up

[MichaelChirico]

@slowkow any general comments on API / expected behavior / style are welcome at any time

[slowkow]

Thanks for working on this! I think I might prefer text_repel() instead of repel_text().

I don't use base graphics very often, so I'm not sure what the preferred interface would be for someone who does. We might want to ask for comments from someone who does use base when you're ready.

When you're ready, could I please ask if you might be able to share a figure that shows this working?

Does the figure redraw if you resize the window? Like this.

[MichaelChirico]

Does the figure redraw if you resize the window?

I can't imagine it would 🤔

I guess ggplot2 is handling that through its special infra -- resizing the plot window re-induces print.ggplot2, which can re-calculate the plot elements.

This could be captured by a dynamic plotting environment for base (e.g. on shiny), but I am not sure it's possible for unadorned base plots.

[MichaelChirico]

@slowkow I added test-base-plots.R (I'll delete it or migrate it before merge), plan is to do a ton of test plots in there.

[MichaelChirico]

@slowkow now tracking plots in this PR under test-base/figures

[slowkow]

Thanks for working on this! Let me know when you want me to have a deeper look. I can be slow to respond and might need multiple reminders.

[MichaelChirico]

Thanks @slowkow -- IIRC there was an issue where the results are highly volatile. Maybe that's an issue you've seen with the gg version of the algorithm as well. I haven't had time to get back to investigate more...