Hi Ander,
On Tue, Aug 7, 2012 at 2:57 AM, Ander Conselvan de Oliveira <
conselv...@gmail.com> wrote:
> What exactly does this fix? Your commit message doesn't make it that
> obvious.
>
I assumed the problem was obvious. The definition of the word opaque makes
it clear. In frame_button_redraw_handler() the code returns (i.e. does not
draw the buttons) if widget->opaque is set. It should be the other way
around.
>
> Weston commit 010f98b0 added the opaque field to struct widget to "track
> and report [...] opaque regions", i.e., the region of a surface whose alpha
> channel is 1.
I don't think this commit is relevant here. What is relevant is the meaning
of opaque. (since cgit.fd.o is down at the moment I will post some
snippets) For widget_set_transparent() you have
widget_set_transparent(struct widget *widget, int transparent)
{
widget->opaque = !transparent;
}
> It seems to me the frame button code is abusing this field to hide the
> buttons in fullscreen mode.
This may be the case but see window.c~:1300 frame_resize_handler()
if (child->opaque) {
widget->window->opaque_region =
wl_compositor_create_region(display->compositor);
wl_region_add(widget->window->opaque_region,
opaque_margin, opaque_margin,
widget->allocation.width - 2 * opaque_margin,
widget->allocation.height - 2 * opaque_margin);
}
it adds the region if opaque is set.
> IMO, either way you define a visibility value based on the opacity value
> results in improper usage.
>
I did not write this code, I'm just building on top of it while trying to
improve it. If you have a better idea, I would be interested to know what
you think.
Regards,
Scott
_______________________________________________
wayland-devel mailing list
wayland-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/wayland-devel
