More Wayland core support: menus are placed correctly. #123
Conversation
o some stubs implemented; o windows are only created when we know their size.
| { | ||
| resize_handler = handler; | ||
| } | ||
|
|
||
| void set_hide_handler(std::function<void()> const& handler) | ||
| void set_hide_handler(std::function<void(bool visible)> const& handler) |
There was a problem hiding this comment.
So this will result in hide_handler(true) actually making it visible? Confusing!
There was a problem hiding this comment.
I'll rename it
| mir::shell::SurfaceSpecification mods; | ||
| mods.state = mir_window_state_fullscreen; | ||
| if (output) | ||
| if (surface_id.as_value()) |
There was a problem hiding this comment.
surface_id can not have a value? /me raises eyebrow.
There was a problem hiding this comment.
surface_id has a value of 0 until we create a window in the Mir scene, after which it identifies the window. (I'll listen to ideas on how to manage this state better.)
| mir_surface->surface_id = surface_id; | ||
|
|
||
| { | ||
| // The shell isn't guaranteed to respect the requested size |
There was a problem hiding this comment.
Any idea if this is a Wayland policy too?
There was a problem hiding this comment.
No direct knowledge, but I don't see how else things can be done e.g. for a tiling WM.
| @@ -1477,7 +1482,7 @@ class WlSeat | |||
| { | |||
| auto me = reinterpret_cast<WlSeat*>(data); | |||
| auto resource = wl_resource_create(client, &wl_seat_interface, | |||
| std::min(version, 6u), id); | |||
| std::min(version, 6u), id); | |||
There was a problem hiding this comment.
This doesn't seem to match our style guide, of 4-space continuations?
| uint32_t /*flags*/) override | ||
| { | ||
| auto const session = session_for_client(client); | ||
| auto* tmp = wl_resource_get_user_data(parent); | ||
| auto& parent_surface = *static_cast<WlSurface*>(tmp); |
There was a problem hiding this comment.
Is this safe? wl_resource_get_user_data(parent) will be a wayland::Surface* type-erased to void*. Shouldn't that be static_cast<WlSurface*>(static_castwayland::Surface*(tmp)), in order to ensure the correct offset?
| uint32_t /*flags*/) override | ||
| { | ||
| auto const session = session_for_client(client); | ||
| auto* tmp = wl_resource_get_user_data(parent); | ||
| auto& parent_surface = *static_cast<WlSurface*>(tmp); |
| .of_type(mir_window_type_freestyle) | ||
| .of_size(geom::Size{640, 480}) | ||
| .with_buffer_stream(mir_surface.stream_id); | ||
| auto* mir_surface = static_cast<WlSurface*>(tmp); |
There was a problem hiding this comment.
Hm, I know this is existing code, but is it actually safe?
wl_resource_get_user_data(surface) will return a void* pointing to a wayland::Surface; I don't think static_cast-ing directly to a WlSurface* is guaranteed to do the right thing?
Shouldn't it be static_cast<WlSurface*>(static_cast<wayland::Surface*>(tmp))?
There was a problem hiding this comment.
You're right. Fix on the way.
AlanGriffiths
force-pushed
the
better-wayland-support
branch
from
485d8e5
to
d716ce5
Compare
|
Ooops, yes. |
Build succeeded |
Some stubs implemented and windows are only created when we know their size.