Fix opaque frame incognito icon and tabstrip positioning.

Fix opaque frame incognito icon and tabstrip positioning.

This affects both MD and non-MD and is based on the current glass behavior + the
fix for bug 580757.

12345678901234567890123456789012345678901234567890123456789012345678901234567890
* Eliminate kAvatarInnerSpacing.  Replace with the opaque frame equivalent of
  https://codereview.chromium.org/1624803002/ (which unfortunately is more
  complicated due to the split of OpaqueBrowserFrameView and
  OpaqueBrowserFrameViewLayout; I'm inclined to undo that split at some point).
  This deals with the "caption buttons on left" case better than the existing
  code, and avoids excessive space in incognito or MD.
* Position/size the incognito icon the same way in both normal and maximized
  mode for MD.  This matches glass, and is necessary because the MD incognito
  icon is much smaller and so shouldn't be positioned/clipped the same way as in
  non-MD.
* Remove unconditional addition of one pixel before tabstrip leading edge.
  Replace with code to position the tabstrip edges based on
  NonClientBorderThickness() instead of FrameBorderThickness() -- in restored
  windows, the two differ by kClientEdgeThickness (i.e. 1 pixel).  This is more
  correct in maximized mode, where that extra pixel is removed, and more correct
  for the "caption buttons on left" case as well, in which case AFAICT we want
  the mirror behavior.  (I don't know what the existing comment about how the
  two sides differ was supposed to mean; I think it's wrong.)  This results in a
  variety of 1-pixel changes to the tabstrip left and/or right edges,
  necessitating some unit test tweaks.
* Call LayoutIncognitoIcon() unconditionally so we can use it to adjust the
  tabstrip bounds even when there's no incognito icon.  This is all a bit more
  confusing than on glass, where we can just lay out a zero-width incognito icon
  and use it to position the tabstrip in all cases, because of the need to
  support caption buttons on either side (screw you Linux).

BUG=519020
TEST=none

Committed: https://crrev.com/e7e177d81bbe474500ec0c0126a0270f6b54ef53
Cr-Commit-Position: refs/heads/master@{#371443}

[Peter Kasting, 2016-01-23 03:34:25]

Description was changed from

==========
Fix opaque frame incognito icon and tabstrip positioning.

This affects both MD and non-MD and is based on the current glass behavior + the
fix for bug 580757.

*

BUG=519020
TEST=none
==========

to

==========
Fix opaque frame incognito icon and tabstrip positioning.

This affects both MD and non-MD and is based on the current glass behavior + the
fix for bug 580757.

12345678901234567890123456789012345678901234567890123456789012345678901234567890
* Eliminate kAvatarInnerSpacing.  Replace with the opaque frame equivalent of
  https://codereview.chromium.org/1624803002/ (which unfortunately is more
  complicated due to the split of OpaqueBrowserFrameView and
  OpaqueBrowserFrameViewLayout; I'm inclined to undo that split at some point).
  This deals with the "caption buttons on left" case better than the existing
  code, and avoids excessive space in incognito or MD.
* Position/size the incognito icon the same way in both normal and maximized
  mode for MD.  This matches glass, and is necessary because the MD incognito
  icon is much smaller and so shouldn't be positioned/clipped the same way as in
  non-MD.
* Remove unconditional addition of one pixel before tabstrip leading edge.
  Replace with code to position the tabstrip edges based on
  NonClientBorderThickness() instead of FrameBorderThickness() -- in restored
  windows, the two differ by kClientEdgeThickness (i.e. 1 pixel).  This is more
  correct in maximized mode, where that extra pixel is removed, and more correct
  for the "caption buttons on left" case as well, in which case AFAICT we want
  the mirror behavior.  (I don't know what the existing comment about how the
  two sides differ was supposed to mean; I think it's wrong.)  This results in a
  variety of 1-pixel changes to the tabstrip left and/or right edges,
  necessitating some unit test tweaks.
* Call LayoutIncognitoIcon() unconditionally so we can use it to adjust the
  tabstrip bounds even when there's no incognito icon.  This is all a bit more
  confusing than on glass, where we can just lay out a zero-width incognito icon
  and use it to position the tabstrip in all cases, because of the need to
  support caption buttons on either side (screw you Linux).

BUG=519020
TEST=none
==========

[Peter Kasting, 2016-01-23 03:34:25]

pkasting@chromium.org changed reviewers:
+ sky@chromium.org

[sky, 2016-01-25 16:19:06]

LGTM

[Peter Kasting, 2016-01-26 02:54:38]

The CQ bit was checked by pkasting@chromium.org

[commit-bot: I haz the power, 2016-01-26 02:55:15]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1622833002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1622833002/1

[commit-bot: I haz the power, 2016-01-26 03:50:37]

Description was changed from

==========
Fix opaque frame incognito icon and tabstrip positioning.

This affects both MD and non-MD and is based on the current glass behavior + the
fix for bug 580757.

12345678901234567890123456789012345678901234567890123456789012345678901234567890
* Eliminate kAvatarInnerSpacing.  Replace with the opaque frame equivalent of
  https://codereview.chromium.org/1624803002/ (which unfortunately is more
  complicated due to the split of OpaqueBrowserFrameView and
  OpaqueBrowserFrameViewLayout; I'm inclined to undo that split at some point).
  This deals with the "caption buttons on left" case better than the existing
  code, and avoids excessive space in incognito or MD.
* Position/size the incognito icon the same way in both normal and maximized
  mode for MD.  This matches glass, and is necessary because the MD incognito
  icon is much smaller and so shouldn't be positioned/clipped the same way as in
  non-MD.
* Remove unconditional addition of one pixel before tabstrip leading edge.
  Replace with code to position the tabstrip edges based on
  NonClientBorderThickness() instead of FrameBorderThickness() -- in restored
  windows, the two differ by kClientEdgeThickness (i.e. 1 pixel).  This is more
  correct in maximized mode, where that extra pixel is removed, and more correct
  for the "caption buttons on left" case as well, in which case AFAICT we want
  the mirror behavior.  (I don't know what the existing comment about how the
  two sides differ was supposed to mean; I think it's wrong.)  This results in a
  variety of 1-pixel changes to the tabstrip left and/or right edges,
  necessitating some unit test tweaks.
* Call LayoutIncognitoIcon() unconditionally so we can use it to adjust the
  tabstrip bounds even when there's no incognito icon.  This is all a bit more
  confusing than on glass, where we can just lay out a zero-width incognito icon
  and use it to position the tabstrip in all cases, because of the need to
  support caption buttons on either side (screw you Linux).

BUG=519020
TEST=none
==========

to

==========
Fix opaque frame incognito icon and tabstrip positioning.

This affects both MD and non-MD and is based on the current glass behavior + the
fix for bug 580757.

12345678901234567890123456789012345678901234567890123456789012345678901234567890
* Eliminate kAvatarInnerSpacing.  Replace with the opaque frame equivalent of
  https://codereview.chromium.org/1624803002/ (which unfortunately is more
  complicated due to the split of OpaqueBrowserFrameView and
  OpaqueBrowserFrameViewLayout; I'm inclined to undo that split at some point).
  This deals with the "caption buttons on left" case better than the existing
  code, and avoids excessive space in incognito or MD.
* Position/size the incognito icon the same way in both normal and maximized
  mode for MD.  This matches glass, and is necessary because the MD incognito
  icon is much smaller and so shouldn't be positioned/clipped the same way as in
  non-MD.
* Remove unconditional addition of one pixel before tabstrip leading edge.
  Replace with code to position the tabstrip edges based on
  NonClientBorderThickness() instead of FrameBorderThickness() -- in restored
  windows, the two differ by kClientEdgeThickness (i.e. 1 pixel).  This is more
  correct in maximized mode, where that extra pixel is removed, and more correct
  for the "caption buttons on left" case as well, in which case AFAICT we want
  the mirror behavior.  (I don't know what the existing comment about how the
  two sides differ was supposed to mean; I think it's wrong.)  This results in a
  variety of 1-pixel changes to the tabstrip left and/or right edges,
  necessitating some unit test tweaks.
* Call LayoutIncognitoIcon() unconditionally so we can use it to adjust the
  tabstrip bounds even when there's no incognito icon.  This is all a bit more
  confusing than on glass, where we can just lay out a zero-width incognito icon
  and use it to position the tabstrip in all cases, because of the need to
  support caption buttons on either side (screw you Linux).

BUG=519020
TEST=none
==========

[commit-bot: I haz the power, 2016-01-26 03:50:38]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-01-26 03:52:07]

Description was changed from

==========
Fix opaque frame incognito icon and tabstrip positioning.

This affects both MD and non-MD and is based on the current glass behavior + the
fix for bug 580757.

12345678901234567890123456789012345678901234567890123456789012345678901234567890
* Eliminate kAvatarInnerSpacing.  Replace with the opaque frame equivalent of
  https://codereview.chromium.org/1624803002/ (which unfortunately is more
  complicated due to the split of OpaqueBrowserFrameView and
  OpaqueBrowserFrameViewLayout; I'm inclined to undo that split at some point).
  This deals with the "caption buttons on left" case better than the existing
  code, and avoids excessive space in incognito or MD.
* Position/size the incognito icon the same way in both normal and maximized
  mode for MD.  This matches glass, and is necessary because the MD incognito
  icon is much smaller and so shouldn't be positioned/clipped the same way as in
  non-MD.
* Remove unconditional addition of one pixel before tabstrip leading edge.
  Replace with code to position the tabstrip edges based on
  NonClientBorderThickness() instead of FrameBorderThickness() -- in restored
  windows, the two differ by kClientEdgeThickness (i.e. 1 pixel).  This is more
  correct in maximized mode, where that extra pixel is removed, and more correct
  for the "caption buttons on left" case as well, in which case AFAICT we want
  the mirror behavior.  (I don't know what the existing comment about how the
  two sides differ was supposed to mean; I think it's wrong.)  This results in a
  variety of 1-pixel changes to the tabstrip left and/or right edges,
  necessitating some unit test tweaks.
* Call LayoutIncognitoIcon() unconditionally so we can use it to adjust the
  tabstrip bounds even when there's no incognito icon.  This is all a bit more
  confusing than on glass, where we can just lay out a zero-width incognito icon
  and use it to position the tabstrip in all cases, because of the need to
  support caption buttons on either side (screw you Linux).

BUG=519020
TEST=none
==========

to

==========
Fix opaque frame incognito icon and tabstrip positioning.

This affects both MD and non-MD and is based on the current glass behavior + the
fix for bug 580757.

12345678901234567890123456789012345678901234567890123456789012345678901234567890
* Eliminate kAvatarInnerSpacing.  Replace with the opaque frame equivalent of
  https://codereview.chromium.org/1624803002/ (which unfortunately is more
  complicated due to the split of OpaqueBrowserFrameView and
  OpaqueBrowserFrameViewLayout; I'm inclined to undo that split at some point).
  This deals with the "caption buttons on left" case better than the existing
  code, and avoids excessive space in incognito or MD.
* Position/size the incognito icon the same way in both normal and maximized
  mode for MD.  This matches glass, and is necessary because the MD incognito
  icon is much smaller and so shouldn't be positioned/clipped the same way as in
  non-MD.
* Remove unconditional addition of one pixel before tabstrip leading edge.
  Replace with code to position the tabstrip edges based on
  NonClientBorderThickness() instead of FrameBorderThickness() -- in restored
  windows, the two differ by kClientEdgeThickness (i.e. 1 pixel).  This is more
  correct in maximized mode, where that extra pixel is removed, and more correct
  for the "caption buttons on left" case as well, in which case AFAICT we want
  the mirror behavior.  (I don't know what the existing comment about how the
  two sides differ was supposed to mean; I think it's wrong.)  This results in a
  variety of 1-pixel changes to the tabstrip left and/or right edges,
  necessitating some unit test tweaks.
* Call LayoutIncognitoIcon() unconditionally so we can use it to adjust the
  tabstrip bounds even when there's no incognito icon.  This is all a bit more
  confusing than on glass, where we can just lay out a zero-width incognito icon
  and use it to position the tabstrip in all cases, because of the need to
  support caption buttons on either side (screw you Linux).

BUG=519020
TEST=none

Committed: https://crrev.com/e7e177d81bbe474500ec0c0126a0270f6b54ef53
Cr-Commit-Position: refs/heads/master@{#371443}
==========

[commit-bot: I haz the power, 2016-01-26 03:52:07]

Patchset 1 (id:??) landed as
https://crrev.com/e7e177d81bbe474500ec0c0126a0270f6b54ef53
Cr-Commit-Position: refs/heads/master@{#371443}