Set Bqplot*Modes to observe only one selected state in BrushSelector

[dhomeier]

Description

Pushing this as a test to avoid possible race conditions in observing both selected_x and selected_y in the BrushSelector interaction that have been speculated to cause unwanted size changes of the selection in #418.

Since the reported bug is non-deterministic and I have not been able to reproduce it at all, I have no idea if this change will help, just keeping it here for further investigation.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (307a11d) 86.63% compared to head (493fada) 86.62%.
Report is 6 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #419      +/-   ##
==========================================
- Coverage   86.63%   86.62%   -0.02%     
==========================================
  Files          89       89              
  Lines        5163     5159       -4     
==========================================
- Hits         4473     4469       -4     
  Misses        690      690              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[pllim]

Won't all the calls to interact.selected_x and interact.selected_y also need to be changed?

[dhomeier]

selected includes both, so any change to an element of either one should alert the observe, but for working with the coordinates using selected_[xy] should just function the same – unless I misunderstand something in the way _set_selected_xy works. But it does work as far as I could see.
https://github.com/bqplot/bqplot/blob/c186fcdbf3c5455c91cb750ab5a3f94accefda81/bqplot/interacts.py#L365-L369

[pllim]

I won't have time for to test this week. Try Cubeviz example notebook if you cannot wait. I could reproduce it there (without this patch) but not consistently.

[dhomeier]

Tried on 3 different macOS and Python versions now, with notebook, jupyterlab and standalone for Cubeviz and the original Imviz example, and never saw as much as a flicker in size – will have to wait then.

[pllim]

@cshanahan1 , since you most recently ran into the "disappearing subset" problem, do you want to see if this PR helps? Given how non-deterministic it is, better have multiple devs confirm.

[pllim]

Maaaybe this fixes some of the problem, but it does not prevent the case where user is dragging something around very fast with repeated down/up clicks. The cursor can accidentally be right at the edge, where the "resize" mode would activate. Then when user intends to move, the app thinks user is resizing, resulting in the shape getting very big or small.

[pllim]

Despite #419 (comment) , listening to just one thing is better than having race condition, right? Should you take this out of draft?

[dhomeier]

Cursor being too close to the edge so it does switch into resize mode seems out of scope for the race condition issue.
But I am not sure I can follow re. the rapid up/down clicks – maybe I need to test this with an actual mouse; with the trackpad there is always some pause before I can lock into drag mode with a new tap.
Anyway if there is at least some evidence this can help, I am changing status now.

[astrofrog]

Seems sensible, thanks!