Re: [Ayatana] No "application bucket" needed

[Luke Morton <luke.morton@xxxxxxxxxxxxxxxx>]

On Tue, 2010-05-18 at 01:49 +0200, Frederik Nnaji wrote: 
> On Tue, May 18, 2010 at 01:10, Luke Morton
> <luke.morton@xxxxxxxxxxxxxxxx> wrote:
>         In the process of writing this, I realised the problem I have
>         with
>         applications closing to the tray is that it makes the
>         consequences of
>         closing windows inconsistent.
>         
>         * Closing the only window for a non-tray application causes
>         the
>         application to quit.
>         * Closing the only window for a tray application does not
>         cause the
>         application to quit.
>         * Most applications are not tray applications so their
>         non-quitting
>         behaviour is inconsistent with the majority.
>         
>         Consider: if you've just opened an application that you've
>         never used
>         before, what would you expect to happen if you closed its
>         window? 
> 
> yes, finally someone!
> 
> 
>         So I think the thing that causes usability problems is
>         actually
>         inconsistent exiting behaviour. 
> 
> Behaviour? This means client-side, not WM, correct?

Yes. The WM emits a signal that the client (application) responds to.
The client (application) can then choose what to do with that signal.

>         If applications never exited when their
>         last window was closed, this wouldn't be a problem.
>         (Incidentally, I
>         think this is the approach Mac OS X takes.) 
> 
> their close button is not red IIRC.
>  
>         
>         > Hitting "close" on one web browser window doesn't terminate
>         the
>         > web-browser process, and the other windows associated with
>         it.
>         
>         
>         It does if its the only window. 
> 
> wow, sometimes it will, sometimes it won't, it makes up for its
> totally inconsistent behaviour with asking the user additional
> questions on exit..
> We all know that's a trace of suboptimal design consistency.
>