Partially missing line on time axis, regression in 4.3.1 about chart.js

[LeeLenaleee]

Yeah its a bug fix so that would sound good to me

[etimberg]

Interesting, this only happens if the min is set on the axis 🤔

@stockiNail I'm guessing this is somehow related to https://github.com/chartjs/Chart.js/pull/11377/files

[stockiNail]

@etimberg @StephanTLavavej I'll have a look.

EDIT: The issue is related to line.element draw where start and count are related to total amount of points and not the subset.

[stockiNail]

@etimberg @LeeLenaleee PR #11432

[StephanTLavavej]

If it makes the investigation easier, I've simplified the repro to a plain linear axis, instead of needing a time axis. The new repro: https://codepen.io/StephanTLavavej/pen/QWJzqWg

[ErikDorstel]

Based on "https://codepen.io/StephanTLavavej/pen/QWJzqWg" the line is not displayed when:

(xmin>4) && (x<(xmin*2)-4)

[stockiNail]

@etimberg @StephanTLavavej @ErikDorstel I have seen where the issue is. To fix #11273 the line dataset element is fill with only the points to show to calculate correctly the control points.
Unfortunately the draw method (but also others to calculate the paths and segments) are invoked passing the start index of the points which is related to the whole amount of points, instead the points in the line element are always starting from.
I need a bit of time to find out the best fix to solve both issues.

[elinake]

I have same issue: https://stackblitz.com/edit/stackblitz-starters-ngv2t8?file=src%2Fapp%2Fapp.component.ts

Bug happens with both wheel and drag and on all axis modes. I have linear scales and no limits. Downgrading to Chartjs 4.3.0 fixed it.

I also noticed that same issue happens in chartjs-plugin-zoom Click-to-zoom sample (only sample with linear scale): https://www.chartjs.org/chartjs-plugin-zoom/latest/samples/wheel/click-zoom.html, even though it says be using 4.2.1.

[etimberg]

@elinake thanks! Under the hood the zoom plugin changes the scale limits to do the zooming so I could see it breaking once a zoom starts

[stockiNail]

@etimberg I think the best thing could be to revert the previous PR to solve this issue. Tee other issue (no blocking) needs more investigation, in my opinion. What do you think? if agree, I'll prepare the PR.

[etimberg]

Sure, lets revert the PR. @LeeLenaleee what do you think about making that into a v4.3.3?