Polar 2.0 open items

[etpinard]

Followup issue of #2200 and the on-par polar milestone.

---

Features we could easily add

• radial and angular axis spikes
• support for radialaxis.side: 'center' to display radial axis labels along invisible radial axis lines
• radial axis hole (similar to pie hole, mpl example) - done in Add polar.hole #2977
• support square grid lines for category angular axes example - done in Polar polygon grids #2739
• some way to snap to NSEW corners during angular drag
• way to activate radial (and possibly angular down the road) autorange on double click (i.e. implement doubleClick: 'autosize' config option
• radial (and angular?) axis zeroline
• polar (r,theta) data-referenced annotations
• trace-specified and radial-axis calendars
• make polar respect gd._context.showAxisDragHandles?

Potential features that require some thought

• How to replace legacy polar's area trace type? I'd vote for replacing it with polarbar or wedge. Most of the bar logic could be reused here. Moved to -> Polar 2.0 replacement for legacy "area" type #2810 Done in -> Add barpolar traces #2954
• How to replace legacy polar's (r,t) bar traces? I'd vote simply removing them, unless someone out there finds them useful.
• What to do with r < 0 values? In Polar 2.0 #2200, they are simply dismissed, but perhaps we should add an option to plot them on the opposite side of the origin. See Polar 2.0 #2200 (comment)
• Date angular axes. See examples in Polar 2.0 #2200 (comment)) and some considerations in the last part of 1e81039#r26653662 Moved to --> Add support for date angular axes #4379
• Should we allow angularaxis.period to be less then the number of categories on categorical angular axes? See Polar 2.0 #2200 (comment)
• Maybe add r0, dr, theta0 and dtheta. For the theta pair, it might be tricky to find a correct default the everyone likes. Done in -> Better polar setConvert + a few misc polar touch ups #2895
• Add another zoombox mode? Perhaps a pixel-based zoom using polar.x, polar.y, polar.zoom scalars to describe it? See Polar 2.0 #2200 (comment)
• scatterpolar (r,theta?) error bars
• Properly clip scatterpolargl line segments

[etpinard]

• make polar subplot drag tests more robust. They currently fail intermittently on CI. They are now skipped on CI as of 8bebc2e. Replicating the issue is hard. I wasn't able to replicate it on my laptop. Even on CI, you need to run all the suites at once to sometime see those tests fail. See commits 40f60de

---

Update: adding that @flaky tag seems to have fixed this pretty well.

[etpinard]

To my surprise, legacy polar charts were able to set the angular range:

https://community.plot.ly/t/custom-angular-range-in-scatterpolar/8676

🆘 We should implement that in polar 2.0 before 🔪 legacy polar charts in v2 🆘

[etpinard]

From @alexcjohnson 's #2739 (comment)

• we should have a mode in scatterpolar to join the endpoints together so you don't need to repeat the initial point when the data are meant to be periodic. trace.connectends perhaps, to mirror trace.connectgaps? Moved to -> Add 'connectends' attribute to scatterpolar lines #3073

[alexcjohnson]

@etpinard a request has come in for radial axis hole (third checkbox above) and legacy area type replacement (first bullet point). hole is straightforward enough that I think this checkbox will suffice to track it, but I'll make a new issue for area so we can hash out the API without overrunning this thread.

[etpinard]

, but I'll make a new issue for area so we can hash out the API without overrunning this thread.

Sounds good!