379316 - [FIX]rgba(0,0,255,4.3) renders with alpha of 0.3 and triggers "ASSERTION: out of range"

Status: RESOLVED FIXED

(Core :: CSS Parsing and Computation, defect, P1)

Description

[Jesse Ruderman]

Attachment: testcase

<div style="color: rgba(0,0,255,4.3)">alpha: 4.3 !?</div>

triggers an assertion:

###!!! ASSERTION: out of range: '0.0 <= aAlpha && aAlpha <= 1.0', file /Users/jruderman/trunk/mozilla/layout/style/nsStyleUtil.h, line 87

I'm also not sure Firefox's rendering is correct.  It renders as if alpha were 0.3, which seems wrong to me, but I don't see anything in http://www.w3.org/TR/css3-color/ that specifies how it should render.  (Two reasonable answers would be "clamp the alpha value to 1" and "ignore the entire rgba as invalid".)

Comment 1

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

http://www.w3.org/TR/css3-color/#alphavaluedt says:

# Any values outside the range 0.0 (fully transparent) to 1.0 (fully opaque)
# will be clamped to this range.

Comment 2

[Boris Zbarsky [:bzbarsky]]

This is a regression from bug 372782.  We do the clamping too late, and nsStyleUtil::FloatToColorComponent automatically takes the remainder mod 255 on return, since the return value is a PRUint8.

Comment 3

[Boris Zbarsky [:bzbarsky]]

Attachment: Fix

Comment 4

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Comment on attachment 263322 [details] [diff] [review]
Fix

r+sr=dbaron.  Thanks for fixing this.

Comment 5

[Boris Zbarsky [:bzbarsky]]

Checked in.

Comment 6

[Robert O'Callahan (:roc) (email my personal email if necessary)]

The last line of the testcase fails on Mac with subpixel anti-aliasing (see bug 379786). I'm not sure how to fix that bug; in the meantime what should we do? Mark the test random on Mac?

Comment 7

[Jesse Ruderman]

How about changing the reftest for this bug to test opacity against opacity and rgba against rgba?

Comment 8

[Boris Zbarsky [:bzbarsky]]

The last line of the test is a control to test that rgba() works at all.  It needs to be tested against opacity.

So we should probably factor it out into a separate test and make that test random on Mac.  I'll do that.

Comment 9

[Boris Zbarsky [:bzbarsky]]

Did that.  That one line is now marked as random on cocoa, pending bug 379786.  Once thats fixed, we should undo it.