lp:~andreas-pokorny/mir/fix-1580774

Branch merges

Merged into lp:mir at revision 3511

Daniel van Vugt: Abstain on 2016-05-17

Cemil Azizoglu (community): Approve on 2016-05-16

Mir CI Bot: Approve (continuous-integration) on 2016-05-16

Kevin DuBois (community): Approve on 2016-05-16

Andreas Pokorny (community): Needs Information on 2016-05-16

Diff: 182 lines (+146/-6)

2 files modified

src/server/scene/mediating_display_changer.cpp (+5/-6)
tests/unit-tests/scene/test_mediating_display_changer.cpp (+141/-0)

api

Propose for merging

Related source package recipes

No recipes using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Bug #1580774: mouse is getting stuck on a phantom edge

High

Fix Released

Link a bug report

Related blueprints

3508. By Andreas Pokorny on 2016-05-16

Include used flag in calculation of input rectangles

3507. By Andreas Pokorny on 2016-05-15

Use the connected status of the display output to calculate input rectangles

The virtual display is powered and available but not connected on startup, hence it would be accounted as reachable display area. The mouse pointer would be allowed to leave the first output, which leads to a termination of input events sent to the nested session.

3506. By Andreas Pokorny on 2016-05-11

Fix not-really-resampling-but-pretending case in InputConsumer

The input resampling batches together InputMessages that up to one relevant state change and fall into a certain time frame. When deciding wether the last InputMessage must be interpolated with the upcoming InputMessage, the previous code used to just add the last coords of the last InputMessage a second time. That makes the cursor move a bit faster than it actually should.

This change does not add linear interpolation of relative mouse data, but only fixes the duplication of the last mouse event. If we turn linear interpolation on for mouse pointers too we need code similar to the one that already exists for x and y axis. Additionally we must subtract the result from the upcoming event...
.

Approved by mir-ci-bot, Alan Griffiths, Chris Halse Rogers, Cemil Azizoglu.

3505. By Alberto Aguirre on 2016-05-11

Add a simple client startup performance test

Add test to ensure the required time for a client to setup a connection,
create a surface and swap a buffer does not exceed the expected threshold.
 .

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

3504. By Daniel van Vugt on 2016-05-11

mesa-kms: When moving the hardware cursor, don't give up and crash if
it fails. I've seen 'Permission denined' occur and crash the server when
VT switching (LP: #1579630), but more generally the hardware cursor
(once initialized) is not an important enough feature to crash the
server for. If any part of it fails, just write a warning to the log
instead.
. Fixes: https://bugs.launchpad.net/bugs/1579630.

Approved by Cemil Azizoglu, Kevin DuBois, mir-ci-bot.

3503. By Daniel van Vugt on 2016-05-10

Add a compositor performance test, which is actually a regression test
for LP: #1563287.

Approved by mir-ci-bot, Kevin DuBois.

3502. By Kevin DuBois on 2016-05-10

nbs api: change mir_presentation_chain_allocate_buffer() to mir_connection_allocate_buffer(). As discussed a few weeks ago, allocating MirBuffers independently of MirPresentationChains makes more sense.

Buffers can change chains, and this opens up the possibility of mir_screencast_fill_buffer(MirScreencast*, MirBuffer*) for more flexible screencasting (useful in WiDi/encoding applications)

Also changes mir_buffer_callback to
typedef void (*mir_buffer_callback)(MirBuffer*, void* context);.

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

3501. By Alan Griffiths on 2016-05-09

platform: change the default behaviour of fatal_error() to abort(). Fixes: https://bugs.launchpad.net/bugs/1577106.

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

3500. By Alan Griffiths on 2016-05-09

common: Improve mir::optional_value<> const propagation & copying.

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

3499. By Chris Halse Rogers on 2016-05-06

Add a couple of extra errors to the list that mg::egl_error understands.

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