587106 - canvas strokeStyle getter mishandles opacity values smaller than 0.098

Status: RESOLVED FIXED

(Core :: Graphics: Canvas2D, defect, P1)

Description

[donald duck]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.2.8) Gecko/20100722 Firefox/3.6.8 (.NET CLR 3.5.30729)
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.2.8) Gecko/20100722 Firefox/3.6.8 (.NET CLR 3.5.30729)

In the canvas element, when setting a strokeStyle with opacity (alpha-value) lower than 0.1, is interpreted incorrectly. For example 0.0666 becomes 0.666, or 0.09 becomes 0.9
This messes up smooth fade-in effects.


Reproducible: Always

Steps to Reproduce:
- Run this script:

<html>
<head>
<script type="text/javascript">
function go(){
  var ctx=document.getElementById('canv').getContext('2d');
  ctx.strokeStyle = 'rgba(0,0,0, 0.09)';
  alert(ctx.strokeStyle);  //FireFox output: (about) 0.9, instead of 0.09
}
</script>
</head><body onload="go()">
<canvas id=canv></canvas>
</body></html>

Actual Results:  
Alert window says "rgba(0, 0, 0, 0.9016)".

Expected Results:  
"rgba(0, 0, 0, 0.09)"  (An alpha-value 10 times smaller).

Opera, Chrome and Safari handle this correctly.

Comment 1

[donald duck]

Attachment: test html page / script

Comment 2

[Nickolay_Ponomarev]

I can reproduce with Mozilla/5.0 (Macintosh; Intel Mac OS X 10.5; rv:2.0b6pre) Gecko/20100904 Firefox/4.0b6pre. Seems to be canvas-specific.

Comment 3

[Boris Zbarsky [:bzbarsky]]

This is a bug in the strokeStyle getter.  It does:

971         PRUint32 alpha = NS_GET_A(aColor) * 100000 / 255;
972         CopyUTF8toUTF16(nsPrintfCString(100, "rgba(%d, %d, %d, 0.%d)",
973                                         NS_GET_R(aColor),
974                                         NS_GET_G(aColor),
975                                         NS_GET_B(aColor),
976                                         alpha),
977                         aStr);

In particular, this will give the wrong answer for any alpha channel value less than or equal to 26 (which corresponds to a floating-point alpha of 0.098 or less).

Comment 4

[Boris Zbarsky [:bzbarsky]]

The right solution is probably to stop using an nsPrintfCString for this.  In fact, I wish we could just reuse the nsCSSValue stringification code here, but that uses "transparent" for NS_RGBA(0,0,0,), which we may not want here.  If we didn't have to do the #rrggbb thing for the alpha == 1 case, we could at least factor out the nsCSSValue code....

Comment 5

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Why do we even -have- that code in there.. what's wrong with just a %f with a precision!

Comment 6

[Nickolay_Ponomarev]

It's from bug 329593, too bad I didn't decide to fix it properly then:

http://bonsai.mozilla.org/cvsblame.cgi?file=mozilla/content/canvas/src/nsCanvasRenderingContext2D.cpp&rev=1.128#570

570 // "%0.5f" in nsPrintfCString would use the locale-specific
571 // decimal separator. That's why we have to do this:

Comment 7

[Boris Zbarsky [:bzbarsky]]

Attachment: Fix

Comment 8

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Comment on attachment 472634 [details] [diff] [review]
Fix

Looks great, thanks -- sorry for the slow review!

Comment 9

[Boris Zbarsky [:bzbarsky]]

Pushed http://hg.mozilla.org/mozilla-central/rev/bcff46a05021

And then bustage fixes:
  http://hg.mozilla.org/mozilla-central/rev/5eb5819ad5bd
  http://hg.mozilla.org/mozilla-central/rev/39db042d6737