Make Chromium windows not hide auto-hide taskbars....

Make Chromium windows not hide auto-hide taskbars.

There are a few caveats here:

* On Aero glass, if the auto-hide taskbar is at the top of the screen, we get one row of nonclient pixels along the bottom of the screen (not too noticeable for light-colored pages, looks a bit odd with a dark page).  I can't find a way around this.

* Switching between fullscreen and normal mode can leave things a bit confused until you click another app and then reactivate Chromium.  This seems to happen with other applications too (e.g. Firefox fullscreen mode) so I'm not too worried.

* Chromium does not deal well with toggling the taskbar's auto-hide setting (or, I think, its position?) unless you restore and remaximize the window.  I tried to fix this via modified handling of WM_SETTINGCHANGE but only made things worse and so gave up.

BUG=20
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=10845

[Peter Kasting, 2009-03-02 23:23:46]

This patch will be out-of-date once Ben lands his frame reorg, but it's
reviewable as-is; I'll update at that time.

Ben, you're a reviewer since this is mostly your code (I think); Amit, you've
been kept in the loop of what I've been doing as I've tried various things.  (I
eliminated the WM_GETMINMAXINFO handling since I didn't actually need it, and in
the process made the code simpler and eliminated the glass-not-drawing problem. 
Yay!)

[amit, 2009-03-03 19:13:43]

very nice!

http://codereview.chromium.org/28338/diff/1/6
File chrome/browser/views/frame/aero_glass_frame.cc (right):

http://codereview.chromium.org/28338/diff/1/6#newcode156
Line 156: reinterpret_cast<RECT*>(l_param);
Nit: in both cases this evaluates to:
RECT* client_rect = reinterpret_cast<RECT*>(l_param);

http://codereview.chromium.org/28338/diff/1/6#newcode178
Line 178: // This is visibly ugly, but there seems to be no better solution.
Move this comments 3 lines lower?

http://codereview.chromium.org/28338/diff/1/9
File chrome/views/custom_frame_window.cc (left):

http://codereview.chromium.org/28338/diff/1/9#oldcode1273
Line 1273: if (IsMaximized()) {
I think this was done to clip out the borders when a window is maximized. This
is so that the borders of a maximized window are not visible on the other
monitors. Is this case covered in the new set of changes?

http://codereview.chromium.org/28338/diff/1/9
File chrome/views/custom_frame_window.cc (right):

http://codereview.chromium.org/28338/diff/1/9#newcode1014
Line 1014: client_rect->bottom -= border_thickness;
InflateRect(client_rect, -border_thickness)

[Peter Kasting, 2009-03-03 22:06:03]

New snap uploaded that syncs with Ben's changes.

http://codereview.chromium.org/28338/diff/1/6
File chrome/browser/views/frame/aero_glass_frame.cc (right):

http://codereview.chromium.org/28338/diff/1/6#newcode156
Line 156: reinterpret_cast<RECT*>(l_param);
On 2009/03/03 19:13:43, amit wrote:
> Nit: in both cases this evaluates to:
> RECT* client_rect = reinterpret_cast<RECT*>(l_param);

True, but this way is more obvious and (in theory) more future-proof.  (In
practice, Microsoft could never change this struct's layout...)

http://codereview.chromium.org/28338/diff/1/6#newcode178
Line 178: // This is visibly ugly, but there seems to be no better solution.
On 2009/03/03 19:13:43, amit wrote:
> Move this comments 3 lines lower?

Sure

http://codereview.chromium.org/28338/diff/1/9
File chrome/views/custom_frame_window.cc (left):

http://codereview.chromium.org/28338/diff/1/9#oldcode1273
Line 1273: if (IsMaximized()) {
On 2009/03/03 19:13:43, amit wrote:
> I think this was done to clip out the borders when a window is maximized. This
> is so that the borders of a maximized window are not visible on the other
> monitors. Is this case covered in the new set of changes?

Yes, the opaque frame GetWindowMask() function has changed.

http://codereview.chromium.org/28338/diff/1/9
File chrome/views/custom_frame_window.cc (right):

http://codereview.chromium.org/28338/diff/1/9#newcode1014
Line 1014: client_rect->bottom -= border_thickness;
On 2009/03/03 19:13:43, amit wrote:
> InflateRect(client_rect, -border_thickness)

OK

[amit, 2009-03-03 22:18:08]

lgtm

[Peter Kasting, 2009-03-04 00:39:11]

BTW, because of tree craziness + deadlines, and amit's LGTM, I landed this; but
I will certainly address any comments from Ben in a followup.