Before this commit when using Dunst with multiple displays and
follow mode mouse/keyboard on a new notification (when no other
notifications are shown) a black window flickers up at the screen where
the last notification was shown.
By calling draw() before x_window_show the new window position will be
calculated and the window will be moved there before the window is shown
and thus remove the flickering.
When the display queue had is full and a new notification would come
in, new notification wouldn't get shown until a currently displayed
notification would timeout.
Even if the notification would have been shown on top of the displayed
queue. So e.g. if the displayed queue would have been filled with
"normal" urgency notifications, an incoming "urgent" notification would
have been delayed.
To let those more important notifications through, the tail of displayed
and head of waiting are swapped on every update if necessary.
For notifications without any timeout, ttl can't be positive.
So when, show_age_threshold is active, dunst won't wake up, if there
are only notifications displayed without any timeout.
While the rule's field will get checked for a NULL value, the
notification's field won't get, as it's assumed to be non-NULL.
This is problematic, as there can be some corner cases, where the
notification's field is actually NULL.
Fixes#536
* Move mouse_left/middle/right_click to global section
* Match the enum value style
* Ignore unknow mouse event
* Split copy-paste code into a function
* Fix typo
All XEvents only got logged with their IDs. It makes more sense to write
out the XEvent's name. Also, writing that we received an XEvent and then
logging again, that we ignored it, makes no sense.
As the get_pixbuf_from_file should retrieve a file's content, it
shouldn't fail silently, if the file isn't there.
For codepaths, which don't use the `icon_path` to search for a given
icon, it gives a clearer error message.
When using round corners, the shape_gc in x_win_round_corners never got
freed. This created a memory leak in x11 itself and let the X process
create more load while becoming sluggish in some situations.
From the docs, XFreeGC actually could return an int and could throw a
BadGC error. Neither of them are important to catch in our situation:
- Having a look into the Xlib sourcecode, XFreeGC returns always 1.
- As we don't overwrite the shape_gc or do any other destoying calls,
the BadGC can't happen in this code.
The icon_path has no point for searching an absolute filename. It was
actually meant to avoid the complex lookup algorithm of the icon theme
spec. And the iconname in this case should be actually just the icon's
basename when searching the path.
Concatenating icons like `~/testimg.png` with all icon paths has no
point and also an additional suffix has no point.
As it currently stands notification_init uses xctx.color to initialize
the notification colors which means that it depends on x11 having been
previously initialized and crashes otherwise. This bug was introduced in
04b327f where draw_setup() was after the startup notification was
created.
As a temporary fix the notification creation was moved under draw_setup
but for a long term solution we should look into removing these xctx
dependencies.
According to the glib docs we should use g_source_add_unix_fd rather
than g_source_add_poll. Additionally, this helps us with memory
management since we don't have to handle the allocation of dpy_pollfd.
