1302385 - SVG events on element with float value positions not always working

Status: RESOLVED DUPLICATE of bug 1167959

(Core :: SVG, defect, P3)

Description

[Francois Guerraz]

Attachment: float.html

User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:51.0) Gecko/20100101 Firefox/51.0
Build ID: 20160912030421

Steps to reproduce:

See the attached float.html


Actual results:

Every rectangle should respond to the mouse over event, only one does. Works on Chrome.

Comment 1

[Francois Guerraz]

Attachment: integers.html

working with integer values only doesn't cause this problem (but I can't use it in my project because I'm depending on other libraries that expect floating point values to work).

Comment 2

[Francois Guerraz]

I think it's a precision issue, if at any time in the scale conversion any of the position attributes becomes too small (~<0.0001) it gets truncated and doesn't get "scaled back up" from the DOM's point of view although it gets displayed properly.

Comment 3

[Francois Guerraz]

Hello,

Any progress on that? This is causing major usability issues with libraries like OpenSeaDragon...

F.

Comment 4

[Francois Guerraz]

Whi is this bug still unconfirmed? I have provided a test case...

Comment 5

[Francois Guerraz]

Hi,

I'm sorry to insist but having not a single clue how to start fixing it myself I really need someone to take a look at this problem.
I have provided a very clear test case and it is preventing some "standard" web libraries to work with Firefox.

Please help.
F.

Comment 6

[Robert Longson [:longsonr]]

Your really should read the guidelines https://bugzilla.mozilla.org/page.cgi?id=etiquette.html (prticularly point 2) before commenting any further.

Comment 7

[Tom Klein]

Attachment: Reduced test case (with 1/60 grid points)

I'm getting the same results here as in bug 1167959 - the role of the viewBox transform in that bug is being played by the scale transform on the <g> in this bug.  (I'm also seeing an identity transform on the anonymous child, but I guess that always happens since bug 1458968.)

The testcase here includes the relevant 1/60 grid points discussed in bug 1167959 - the yellow one was the one I hit in a test run.