Re: Would there be a drawback of using the same graphical toolkit on every platform?

[Po Lu]

Arthur Miller <arthur.miller@live.com> writes:

> Is it?

It is.

> Several entire desktops and countless applications implemented with
> Gtk for different OS:s certainly speak in favor of your arguments.

Emacs is more complicated and demanding a program than most of those
desktops and applications.  For instance, they typically don't support
connecting to and displaying windows on multiple displays from the same
process.

> Cetainly it must be problem to Gtk and someone elses incompetence,
> can't be that Emacs is using it in a manner it wasn't supposed to be
> used.

So we should limit our features to what GTK provides?  Just to name a
few, that would entail removing support for multi-TTY, distinguishing
"C-S-u" from "C-u", controlling the size of scroll bars, reliable
control over frame size?

If you want such an experience, by all means use the PGTK build.  But
those features are important to many other people, so we cannot rely on
PGTK.

> So Gtk should be designed as Haiku? Otherwise it is a bug? :)

I'm not, but the problem has nothing to do with how Emacs handles the
toolkit event loop.

> You can either do as you do and consider a framework design to be a
> bug because it does not fit into your wishful expectations

I expect a toolkit which people expect to become the basis of the Emacs
GUI to be designed suitably, which is hardly wishful thinking.

> or you can recognize that the design does not fit a particular
> application in this case Emacs. There is nothing wrong with that, none
> framework is required to fit each and every use case in
> existence. Emacs uses Gtk in a way it is not supposed to and that
> creates some friction.

Whether or not it's suitable for other applications, the friction
between Emacs and GTK cannot be resolved, so that indeed makes GTK
unsuitable as the sole window system for Emacs.

> I don't understand why you need to blame that on Gtk? I am not even
> very fond of Gtk myself, but there is no reason to be unfair.

When a toolkit that is part of the GNU project makes deliberate design
decisions which prevents it from being used to implement Emacs, I think
it is quite right to complain.  Especially when older versions of said
toolkit used to work fine.