Comments on: Window managers and D-Bus http://blogs.gnome.org/metacity/2010/01/16/window-managers-and-d-bus/ "Many window managers are like Marshmallow Froot Loops; Metacity is like Cheerios." Sun, 26 Sep 2010 15:08:40 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Dan Erat http://blogs.gnome.org/metacity/2010/01/16/window-managers-and-d-bus/comment-page-1/#comment-1179 Dan Erat Tue, 19 Jan 2010 20:43:47 +0000 http://blogs.gnome.org/metacity/?p=659#comment-1179 Using D-BUS as an alternate communication channel between window managers and other clients seems like it'd bring some tricky synchronization issues. What happens if a message that a client sends about a window via D-BUS arrives at the WM before the window's CreateNotify event? Using D-BUS as an alternate communication channel between window managers and other clients seems like it’d bring some tricky synchronization issues. What happens if a message that a client sends about a window via D-BUS arrives at the WM before the window’s CreateNotify event?

]]> By: Thomas Thurman http://blogs.gnome.org/metacity/2010/01/16/window-managers-and-d-bus/comment-page-1/#comment-1176 Thomas Thurman Sun, 17 Jan 2010 19:56:27 +0000 http://blogs.gnome.org/metacity/?p=659#comment-1176 Right, which is another reason why it would be unreasonable just to drop EWMH and go with D-Bus. Right, which is another reason why it would be unreasonable just to drop EWMH and go with D-Bus.

]]> By: Simon http://blogs.gnome.org/metacity/2010/01/16/window-managers-and-d-bus/comment-page-1/#comment-1175 Simon Sun, 17 Jan 2010 19:53:38 +0000 http://blogs.gnome.org/metacity/?p=659#comment-1175 The existing EWMH system works perfectly with remote X, which I use extensively. D-Bus, not so well, assuming it's even available on the remote machine. Which it frequently isn't, since in my case the remote machines are often AIX or HP-UX servers... The existing EWMH system works perfectly with remote X, which I use extensively. D-Bus, not so well, assuming it’s even available on the remote machine. Which it frequently isn’t, since in my case the remote machines are often AIX or HP-UX servers…

]]> By: Screwtape http://blogs.gnome.org/metacity/2010/01/16/window-managers-and-d-bus/comment-page-1/#comment-1173 Screwtape Sat, 16 Jan 2010 08:35:36 +0000 http://blogs.gnome.org/metacity/?p=659#comment-1173 Long time no see! Welcome back, Thomas! I guess pros of D-Bus are that people familiar with ordinary app programming should already be familiar with it; the details of X11 are pretty well hidden by GTK+ and Qt these days. Also, D-Bus supports introspection, so you can poke around with a tool like D-Feet in the (inevitable) event of inadequate documentation. The cons would be having to create a sister spec to EWMH, or even a new version that describes D-Bus bindings as well as traditional X11 APIs, and waiting for people to adopt it. Frankly, I think your Perl shim sounds like the best of both worlds - functionality available to D-Bus, without having to write extra code for every EWMH window-manager. Bonus points if it includes libwnck's functionality into the same data model, too. Long time no see! Welcome back, Thomas!

I guess pros of D-Bus are that people familiar with ordinary app programming should already be familiar with it; the details of X11 are pretty well hidden by GTK+ and Qt these days. Also, D-Bus supports introspection, so you can poke around with a tool like D-Feet in the (inevitable) event of inadequate documentation.

The cons would be having to create a sister spec to EWMH, or even a new version that describes D-Bus bindings as well as traditional X11 APIs, and waiting for people to adopt it.

Frankly, I think your Perl shim sounds like the best of both worlds – functionality available to D-Bus, without having to write extra code for every EWMH window-manager. Bonus points if it includes libwnck’s functionality into the same data model, too.

]]>