Closed Bug 992731 Opened 6 years ago Closed 6 years ago

Stop SVGPathElement.getTotalLength from returning NaN

Categories

(Core :: SVG, defect)

x86_64
All
defect
Not set

Tracking

()

VERIFIED FIXED
mozilla32
Tracking Status
firefox30 --- unaffected
firefox31 - affected
firefox32 --- verified

People

(Reporter: alice0775, Assigned: bas.schouten)

References

()

Details

(Keywords: regression, site-compat, testcase)

Attachments

(4 files)

Steps To Reproduce:
1. Open http://cmx.io/edit/sample.html

Actual Results:
a single green line is drawn

And an error shown:
Error: TypeError: Argument 1 of SVGPathElement.getPointAtLength is not a finite floating-point value.
Source File: http://cmx.io/v/0.1/cmx.js
Line: 22654

Expected Results:
Cartoon should be drawn


http://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=2c4c218a7a09&tochange=f853e8ff85ea

Regressed by:
f853e8ff85ea	Bas Schouten — Bug 987077: Correctly deal with a constant acceleration. r=jwatt
Bas, Can you have a look at this?
Flags: needinfo?(bas)
jwatt, can you come up with a minimal testcase for me?
Flags: needinfo?(bas) → needinfo?(jwatt)
Keywords: site-compat
Jonathan, could you have a look here? I can still reproduce it with today aurora. Thanks
Attached image reduced testcase
Flags: needinfo?(jwatt)
Summary: SVG cartoon fails to draw with Error: TypeError: Argument 1 of SVGPathElement.getPointAtLength is not a finite floating-point value. → Stop SVGPathElement.getTotalLength from returning NaN
(I can reproduce in 64-bit Linux nightly builds. OS-->All, version-->Trunk)
OS: Windows 7 → All
Version: 31 Branch → Trunk
Assignee: nobody → bas
Status: NEW → ASSIGNED
Attachment #8431918 - Flags: review?(jwatt)
Attachment #8431918 - Flags: review?(jwatt) → review+
This kind of thing should really have a test case when it's checked in.
I have a testcase in my queue on the Moz2D repo, to land once I do a merge from m-c to moz2d.
https://hg.mozilla.org/mozilla-central/rev/9f56d8be7e42
Status: ASSIGNED → RESOLVED
Closed: 6 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla32
Attached image chrome vs nightly.png
There several differences comparing the drawings in chrome vs nightly 32.0a1 (2014-06-02). Also the values returned by the reduced testcases are different.
Does it actually matter?
Flags: needinfo?(bas)
Thanks, Paul -- I see the same results (much too tiny box around "One Lazy Morning" in Nightly).  Firefox release version matches Chrome, with a normal-sized border-box.

Probably best to track that in a separate bug, though. Mind filing?
Done - bug 1020255
Status: RESOLVED → VERIFIED
Flags: needinfo?(paul.silaghi)
Bas, can we get an uplift request for 31? Thanks.
Too late for 31.
Untracking it.
Depends on: 1068603
Flags: needinfo?(bas)
Depends on: 1116541
You need to log in before you can comment on or make changes to this bug.