feature: adding an is_above method to the SurfaceStack so that compositor authors can write tests to assert the relative z-order of surfaces

[mattkae]

What's new?

• Bumped the project minor version to 20 (new symbols)
• Added AbstractShell::is_above, FocusController::is_above, and SurfaceStack::is_above
• Added methods to the WindowManagementTestHarness to focus a window and query is_above
• Added two MinimalWindowManagerTests that assert points about is_above
• Updated the symbols.map file

[robert-ancell]

Code changes look fine, will need someone else who better knows Mir architecture to clarify if there should be this z_order feature.

[AlanGriffiths]

I'm not convinced that we have to bake in a z-order to our APIs. If we ever start positioning surfaces in 3D, then the z-order will depend on viewpoint(s).

Having said that, we already have:

    void raise(SurfaceSet const& surfaces);
    void swap_z_order(SurfaceSet const& first, SurfaceSet const& second);
    void send_to_back(SurfaceSet const& surfaces);

So, maybe that thought is moot.

Now, the semantics of z_order() are not intuitive.

1. '0' == "at front" OR "expired"OR "not found"
2. layers are not respected (comparing windows from different layers doesn't work)

I think we would do better with an interface for making assertions:

EXPECT_TRUE(shell->assert_that(window1, is_above, window2));

[mattkae]

I think we would do better with an interface for making assertions:

EXPECT_TRUE(shell->assert_that(window1, is_above, window2));

@AlanGriffiths Do you mean this API literally, or would the shell have a method like Shell::is_above(Window const&, Window const&)?

[AlanGriffiths]

@AlanGriffiths Do you mean this API literally, or would the shell have a method like Shell::is_above(Window const&, Window const&)?

The API I suggested is more fluent (reads more like English) and allows some additional predicates to be added without breaking ABI (e.g. shares_layer_with). But I didn't really intend it literally. HTH

[AlanGriffiths]

Just some nitpicking