lp:~mir-team/mir/x11-multi-window

Branch merges

Propose for merging

No branches dependent on this one.

On hold for merging into lp:mir

Mir development team: Pending requested 2017-02-16

Diff: 708 lines (+210/-89)

10 files modified

src/platforms/mesa/server/x11/graphics/display.cpp (+81/-44)
src/platforms/mesa/server/x11/graphics/display.h (+15/-5)
src/platforms/mesa/server/x11/graphics/display_buffer.cpp (+13/-5)
src/platforms/mesa/server/x11/graphics/display_buffer.h (+6/-2)
src/platforms/mesa/server/x11/graphics/display_configuration.cpp (+26/-17)
src/platforms/mesa/server/x11/graphics/display_configuration.h (+8/-8)
src/platforms/mesa/server/x11/graphics/graphics.cpp (+8/-0)
src/platforms/mesa/server/x11/graphics/platform.cpp (+4/-3)
src/platforms/mesa/server/x11/graphics/platform.h (+2/-0)
src/platforms/mesa/server/x11/input/input_platform.cpp (+47/-5)

api

Related source package recipes

No recipes using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Related blueprints

4037. By Nick Dedekind on 2017-02-16

Multiple window support

4036. By Andreas Pokorny on 2017-02-16

Remove android-input

This now also removes the separate thread created for every client window. .

Approved by mir-ci-bot, Daniel van Vugt, Cemil Azizoglu, Brandon Schaefer.

4035. By Daniel van Vugt on 2017-02-16

Add letterboxing/black bars support to the GL renderer, for when
the logical viewport aspect doesn't match that of the screen.

This will be used in future for arbitrary cloning of one monitor
to another (LP: #1639226)
. Fixes: https://bugs.launchpad.net/bugs/1639226.

Approved by Brandon Schaefer, mir-ci-bot.

4034. By Andreas Pokorny on 2017-02-15

mirserver: Use the normal mir connection to send input to clients

Input Events are now sent through the connection and contain a window id to refer to the destination window. So all types of events occur on the client side in the order they were sent from the server. Additionally the conversion between android::InputMessage and MirEvent is avoided. This simplifies adding new event types or extending the content of existing input events. The client code needed only minor changes to not filter out MirInputEvent.
.

Approved by Brandon Schaefer, Kevin DuBois, mir-ci-bot.

4033. By Nick Dedekind on 2017-02-15

Added EDID to output configuration update. (LP: #1660017). Fixes: https://bugs.launchpad.net/bugs/1660017.

Approved by Daniel van Vugt, Kevin DuBois, mir-ci-bot.

4032. By Daniel van Vugt on 2017-02-15

Simplify hardware cursor construction (no parameter required).

Previously hardware cursor implementations were expected to show
on construction with a given image. We now instead hide on construction but show an image immediately after as part of the default configuration policy.

Aside from providing a simpler interface, this is required as a
prerequisite for runtime fallback to the software cursor when the
hardware cursor won't be used (both cursors exist but never more than
one of them shown). This is also a prerequisite to fixing bug 1662760.

Assuming that a hardware cursor might only fail during construction was
a poor design. Going forward we will need to deal with a hardware cursor
that might become unusable or undesirable at any point in its life. Fixes: https://bugs.launchpad.net/bugs/1639226.

Approved by mir-ci-bot, Daniel van Vugt, Chris Halse Rogers.

4031. By Andreas Pokorny on 2017-02-14

client: Add missing event printer for output event and use int RPC Report

The RPC Report now logs each successfully received event.

Approved by mir-ci-bot, Brandon Schaefer, Cemil Azizoglu.

4030. By Cemil Azizoglu on 2017-02-14

Move newly published APIs example code from playground/ to examples/.
.

Approved by Daniel van Vugt, mir-ci-bot.

4029. By Kevin DuBois on 2017-02-14

distinguish the ability to read from a texture from the ability to write to a texture directly by adding the TextureTarget interface.

This is for screencasting in the line of the screencast-to-buffer work. Specifically, the problem comes in the case that the user allocates a ShmBuffer and then tries to write to it via the CompositingScreencast. The ShmBuffer gets uploaded as a texture, and then ScreencastDisplayBuffer will write to that texture, but the pixel content never gets downloaded back to the ShmBuffer. This is different from how gbm_bo and ANativeWindowBuffer binding works, where the way that is bound (via ImageTargetTexture2D) has the ScreencastDisplayBuffer writing directly to the memory backing those two types.

This branch allows us to detect when a user tries to write to a ShmBuffer, which in the current code, will not modify the content of ShmBuffer

.

Approved by mir-ci-bot, Daniel van Vugt, Cemil Azizoglu.

4028. By Daniel van Vugt on 2017-02-14

Fix a strange little bug in the DRM cursor code where:
  hide(); show(); // WORKS
but
  hide(); show(image); // FAILS.

Approved by Chris Halse Rogers, mir-ci-bot.