User Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) Gecko/20100101 Firefox/31.0 (Beta/Release) Build ID: 20140716183446 Steps to reproduce: Open this site: http://kovacsv.hu/errors/pointsonpath.html Actual results: The red dots not follow the path, and the total length value is incorrect. Expected results: The points should be on the path (as in Chrome 36 and Internet Explorer 11). The path total length should be near 448.

Setting product/component to match bug 1024926

Works for me. Does turning hardware acceleration off fix it for you?

(In reply to Robert Longson from comment #2) > Works for me. Does turning hardware acceleration off fix it for you? I updated the link with another example: http://kovacsv.hu/errors/pointsonpath.html My experiences with hardware acceleration: ON: Both examples wrong. OFF: First example seems good, but the second is still wrong.

You should make the second example a different bug as it seems to exhibit different symptoms with regards to HWA.

(In reply to Robert Longson from comment #4) > You should make the second example a different bug as it seems to exhibit > different symptoms with regards to HWA. I separated the two errors. From now, this bug is only about the first error, which occurs only if hardware acceleration is on: http://kovacsv.hu/errors/pointsonpath.html I will file another bug about the second test case.

Returns 0 when (x2, y2) == (x, y). In this case should be around 66.

in FindInflectionApproximationRange cp41.x == cp41.y so *aMin == -INF and *aMax = INF and then everything that follows skips out

Result matches IE and Chrome in Ubuntu 15.10 with Nightly with the patch from bug 1134549 applied, which was recently pushed in https://hg.mozilla.org/mozilla-central/rev/3e73fc0dfb01