tray: allow css styling

[trigg]

tray: add icons to button
tray: fix click logic for button
tray: attach popup menu to button

[trigg]

Switch the eventbox which is not usable via CSS to a button, and account for the mouse-press differences.

Attach popup corner to icon corner, Gtk accounts for orientation issues automatically

[NamorNiradnug]

Hi, thanks for your PR!

As far as I remember GtkButton doesn't handle scroll events while EventBox does, and that's the reason EventBox is used currently. Please consider applying some workaround to keep scrolling above tray items working.

[trigg]

I can't say I know a single system item that uses scroll wheel, got a program I can test on?

[trigg]

From my tests the widget acts the same as before for scroll, but everything I have runs through the GDK_SCROLL_SMOOTH code path... even the discreet scroll wheel.

[soreau]

From my tests the widget acts the same as before for scroll, but everything I have runs through the GDK_SCROLL_SMOOTH code path... even the discreet scroll wheel.

If you read the documentation, you must first consider the value of the direction field to see if the event is GDK_SCROLL_[CARDINAL] or GDK_SCROLL_SMOOTH. If it is the latter, only then can you use the delta fields.

[trigg]

Allow me to rephrase:

All my devices report an ev->direction value of GDK_SCROLL_EVENT.

I am unable to do testing on the alternative codepaths for GDK_SCOLL_* but can't see anything there that would fail when using Gtk::Button in place of Gtk::EventBox.

So:

Each individual 'click' of my mouse wheel calls the signal once with a delta_y of +- 1.5
It takes 4 clicks in a direction before one dbus "Scroll" call is made.

[trigg]

With a bunch of std::cout calls plastered about I've come to a conclusion:

On (at least) my version of GTK Every Single scroll event will pose as Smooth scroll until I remove the flag here.

wf-shell/src/panel/widgets/tray/item.cpp

Line 122 in 7320d40

add_events(Gdk::SCROLL_MASK | Gdk::SMOOTH_SCROLL_MASK);

After this is removed I get GDK_SCROLL_UP and GDK_SCROLL_DOWN on my wheel correctly, and now every 5th click will actually send the signal to the tray program.

So this rules out the previous statement that I couldn't test that code path. I just needed to jump a few more hoops to reach it.

This branch doesn't alter scroll wheel responsiveness on any of my tests.