Serenity: LibWeb/LibGfx: Path filling behaves incorrectly

[linusg]

Just filing an issue for the sake of having one - I haven't actually investigated what's causing this. Might or might not be easy to fix.

canvas-path-quadratic-curve.html:

[alimpfard]

I believe that's because we don't implicitly close paths when trying to fill them.
don't quote me on that though.

[alimpfard]

It's actually kind of odd that firefox and chrome do this; maybe stroke() is supposed to close the path?

adding ctx.closePath() after stroke() makes it looks correct...

[awesomekling]

Some CRC2D ops indeed do implicit path closing!
https://html.spec.whatwg.org/multipage/canvas.html#dom-context-2d-fill

[alimpfard]

Open subpaths must be implicitly closed when being filled

ouch.
we don't really have such a concept.
that's a weird thing to do...as expected from _the web_, nothing is ever straightforward.

[awesomekling]

Let's just do a naive solution for now? Make a copy of the path and call close() on it. Doesn't have to be perfect, just has to work at first :)

[alimpfard]

That would attempt to close the last segment, which is already closed.

unless I'm just remembering how close() works wrong

edit:
what's even going on in Path::close()? it looks like we're trying to close the first "open" path and then immediately exiting?

[awesomekling]

Path::close() is same semantics as CanvasRenderingContext2D.closePath:

Causes the point of the pen to move back to the start of the current sub-path. It tries to draw a straight line from the current point to the start. If the shape has already been closed or has only one point, this function does nothing.

[alimpfard]

What if we had multiple "open subpaths"?

fill() suggests that we have to close _all_ of them, but closePath() seems to suggest that we only close the last one?

[alimpfard]

Yeah, just calling close won't work, it'll try to connect the beginning of the bezier curve to the start point of the other shape:

[awesomekling]

Ahh I see! Yeah I misread the spec, we should indeed implicitly close all open subpaths.

[alimpfard]

I guess I'll hack something up :shrug:

[linusg]

Thanks guys, lovely watching you two figuring this out! :smile: