1259787 - Scaling a line along 1 axis causes it to shift along the other axis

Status: RESOLVED WORKSFORME

(Core :: SVG, defect)

Description

[brendan.d.mcnamara]

Attachment: firefox bug.zip

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_10_5) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/49.0.2623.75 Safari/537.36

Steps to reproduce:

- drawing a border using lines (for the sides) and arcs (for a corner radius).
- draw an arc of radius 2 and try to line it up with the lines from the border


Actual results:

The ends of the arc are not lining up with borders (works in Chrome and Safari)
- image of code and snapshot of page are attached
- seems to be off by half a pixel in either direction


Expected results:

The corner and borders should line up.

NOTE: Code is written in react, hard-coded "2" represents corner radius of 2 in SVG

Comment 1

[Robert Longson [:longsonr]]

We really are going to need a minimal testcase here and not just a set of png images. Please add a single html file demonstrating the issue.

Comment 2

[brendan.d.mcnamara]

Attachment: firefox demo.zip

Sorry about that, I did some more digging and boiled down the issue to this: scaling a line along 1 axis causes it to shift along the other axis. Attached is another set of images and the html code that renders the problem

Comment 3

[Robert Longson [:longsonr]]

Attachment: requestor's testcase

Comment 4

[Loic]

It's an old regression:
http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=f7c89de3ab43&tochange=3f0587ce1774


Matt Woodrow — Bug 788044 - Make transformed frames the reference frame for their display list tree. r=roc

Comment 5

[Marco Perez [:mpk]]

Attachment: multiple minimized testcases

The problem seems to be triggered by using the line transformation as a style property. The direction and amount of the shift depends heavily on the page zoom but can also be influenced by using e.g. a path instead of a line.

Is there a connection between this bug and bug #1265264?

Comment 6

[Robert Longson [:longsonr]]

I don't know. I suggest you establish a regression range in the other bug.

Comment 8

[brendan.d.mcnamara]

I'm updated to version 46.01 on Firefox and this bug continues to exist. Is there any progress on fixing this?

Comment 9

[Robert Longson [:longsonr]]

Seems OK now.