bugfix/20166-columns-gridLines-and-pointPlacement

[MarkusBarstad]

Fixed #20166, columns with pointPlacement had a superfluous tick to the left of the chart, visible with gridLineWidth.

Note: Firefox headless broke at me so this was committed with --no-verify so I could review github-side checks 😬

This could be a messy addition to otherwise clean code but I am unsure if i should:

• Create a variable at the top of the axis.render function which checks for column series with pointPlacement
• Create it closer to the check
• Handle it some other way

---

• To see the specific tasks where the Asana app for GitHub is being used, see below:
  • https://app.asana.com/0/0/1206008636612601

[highsoft-bot]

File size comparison

Sizes for compiled+gzipped (bold) and compiled files.

	master	candidate	difference
highcharts.js	96.0 kB 271.3 kB	96.0 kB 271.4 kB	23 B 47 B
highstock.js	128.7 kB 372.7 kB	128.8 kB 372.7 kB	24 B 47 B
highmaps.js	121.5 kB 349.3 kB	121.5 kB 349.4 kB	39 B 95 B
highcharts-gantt.js	131.0 kB 377.5 kB	131.0 kB 377.6 kB	24 B 47 B
modules/heatmap.js	7.2 kB 19.3 kB	7.2 kB 19.3 kB	11 B 48 B

[highsoft-bot]

Visual test results - Differences found

Found 1 diffing sample(s). Please review the differences.

[bre1470]

Good news: It works here with Firefox ESR (v115).

[bre1470]

Looks good so far. This fixes the second issue example.

Things that should be improved:

• broaden the scope of the fix
• a visual test is needed, this properly can be part of an existing sample

[bre1470]

You should fix this for all cartesian series types. Instead of series.is(... check series.xAxis.

[MarkusBarstad]

Good catch!

It now checks for series.xAxis and unit-test has been altered to check for this as well.

Might not have come out clearly in the issue but this is the only example which I was able to replicate both in jsfiddle as well as utils.

I have however uncovered a similar issue for the bar series, but it does seem to arise from the same conditions (tick[-1] being rendered for cartesian series using pointPlacement)

Should I create a seperate issue for the bar series bug?

[bre1470]

The bar series is cartesian, but it has inverted axis. This needs taken into account and is probably related.
See https://api.highcharts.com/highcharts/chart.inverted

[TorsteinHonsi]

Instead of looking for series with the pointPlacement option, I think we should just check that the tick position is inside the axis. That would be simpler and safer.

For the grid line, we have a force option for the getPlotLinePath function. For some reason it is set to pass for the grid lines. If we comment out that forde parameters, the problematic grid line is not rendered. So I think we should find out why and in what cases we want to force it to render.

For the tick mark we don't seem to have that kind of overflow protection, but we should probably have something similar. Anything that is not between axis.pos and axis.pos + axis.length should not be rendered.

[bre1470]

When running npx gulp test --browsercount 1 there are now 11 instead of 6 (master) tests failing.

[MarkusBarstad]

When running npx gulp test --browsercount 1 there are now 11 instead of 6 (master) tests failing.

Yeah you are right, I think gulp "got stale" or something and did not actually track my changes, allowing a failing commit.

[hubertkozik]

The work will be continued in this PR: #20467 Due to many no-verify commits here, which can cause issues with future bisecting.