Fix glitch in rendering of the stroke at the bottom of the browser toolbar....

Fix glitch in rendering of the stroke at the bottom of the browser toolbar.

The code that rendered this line was doing weird things (adding/removing kClientEdgeThickness from the edges?!) so I removed this and used toolbar_bounds.right() instead of toolbar_bounds.width() which is correct given that toolbar_bounds.x() > 0.

I also got rid of the use of MirroredLeftPointForRect since the flipped version of a horizontal line is... the same horizontal line.

I have also updated the glass frame not to subtract a pixel from the right edge which looks glitchy as well.

http://crbug.com/48134
TEST=see bug


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=53998

[Peter Kasting, 2010-07-27 23:18:38]

Satoshi Matsuzaki created a changelist for this same bug at
http://codereview.chromium.org/3042004 .  Could you look at it?

There are a couple of differences between his patch and yours:
* He still subtracts kClientEdgeThickness on the last argument.  This looks to
me like a bug in your patch.
* He increases the use of MirroredLeftPointForRect() rather than reducing it.  I
believe both your changes will wind up drawing the same thing in that regard,
but I'm not 100% sure.
* He modifies the glass view as well, which looks to me like an omission in your
patch.

If I'm right in all of these, then the change we probably want is something that
looks like his patch except dropping all the MirroredLeftPointForRect() calls.

[Ben Goodger (Google), 2010-07-27 23:35:38]

Updated:

On 2010/07/27 23:18:38, Peter Kasting wrote:
> There are a couple of differences between his patch and yours:
> * He still subtracts kClientEdgeThickness on the last argument.  This looks to
> me like a bug in your patch.

Fixed. Read DrawLine as FillRect.

> * He increases the use of MirroredLeftPointForRect() rather than reducing it. 
I
> believe both your changes will wind up drawing the same thing in that regard,
> but I'm not 100% sure.

Probably. I still removed it because I think it makes the code less readable.

> * He modifies the glass view as well, which looks to me like an omission in
your
> patch.

I updated this to include glass too... and made that code consistent with opaque
(in the process fixing a 1px glitch there).

[Peter Kasting, 2010-07-27 23:43:25]

LGTM with nit.

http://codereview.chromium.org/3056022/diff/8001/9001
File chrome/browser/views/frame/glass_browser_frame_view.cc (right):

http://codereview.chromium.org/3056022/diff/8001/9001#newcode339
chrome/browser/views/frame/glass_browser_frame_view.cc:339:
toolbar_bounds.bottom() - 1,
Nit: These 1s should probably be kClientEdgeThickness.