Web lists-archives.com

Re: GTK_MODULES removal and the future of existing modules




Thank you for the quick response!

I'm not sure anything short of direct access to the widget tree would
suffice for a GTK4 version of Plotinus to provide the functionality it
does today on GTK3.

The problem is that in practice, some of the most important
applications use GTK in an "incomplete" manner. A good example is
LibreOffice: It doesn't use GtkApplication, nor does it have
GActionMaps, or GMenuModels from which actions can be extracted. There
is only a GtkMenuBar widget, which I'm guessing is constructed from
some backend-agnostic model inside the LibreOffice GUI stack.

Plotinus currently tries the obvious things first, looking for
application menus and their models, and falls back to raw widgets if it
doesn't find any. It is this fallback feature that allows Plotinus to
work with software like LibreOffice, and it is software like
LibreOffice where Plotinus is the most useful, since it has hundreds of
menu items.

Of course, I don't expect LibreOffice to move to GTK4 in the short or
even medium term. The major cross-platform applications like Firefox,
LibreOffice, GIMP, and Inkscape have been very reluctant in their
adoption of even GTK3, and likely don't fancy another bumpy upgrade
anytime soon. Still, eventually they might switch, and it makes sense
to anticipate that.

For Plotinus to continue supporting "semi-GTK" applications,
introspection APIs would need to cover the case where there is nothing
but a GtkWindow and a bunch of widgets. The existing D-Bus APIs, for
example, all seem to require at least a GApplication. I don't know
enough about GTK internals to say which alternatives might be feasible.
You mention GIO extension points, a feature I wasn't aware of until
now. How could such an extension point look for Plotinus' use case?
What information would it provide? And would it work with software that
doesn't use GApplication?

Thanks & best regards
Philipp




On Sun, 2018-02-25 at 09:54 +0000, Emmanuele Bassi wrote:
> Hi;
> 
> On Sun, 25 Feb 2018 at 09:18, Philipp Emanuel Weidmann <pew@worldwide
> mann.com> wrote:
> > Greetings,
> > 
> > I am the author of Plotinus[1], a GTK+ module that provides a
> > searchable command palette to GTK+ applications. Recently, it was
> > brought to my attention[2] that module loading has been removed[3]
> > on
> > GTK+ master.
> > 
> > It appears that this change could mean the end for Plotinus and
> > other
> > modules like it. I would be interested to learn:
> > 
> > 1. What is the rationale behind the removal of module loading?
> > 
> 
> The module code was fairly ancient, and was hand rolling things for
> which we have better API down the stack, like GIO extension points,
> which support things like priorities and prerequisites.
> > 2. What will be the first stable release of GTK+ that does not
> > support
> >    modules anymore? Is this GTK+ 4.0+ only, or will support also be
> >    dropped in a 3.0 series release?
> > 
> 
> The GTK 3.x series is frozen, so it won’t be touched. This change is
> for 4.x only.
> 
> > 3. What, if any, alternatives are available/planned for software
> >    like Plotinus that needs to inspect the widget hierarchy of
> > running
> >    applications in order to work?
> > 
> 
> We could be amenable to add an extension point for this, if you
> present a case for it, and explain what kind of requirements you need
> from the toolkit; granting blanket access to the internals of the
> toolkit is not something we’d be happy to provide, but if you have a
> specific domain it should be possible to accommodate your extension.
> 
> Alternatively, we could ensure that all our menus and actions are
> introspectable from the outside, and give you a proper API for
> writing Plotinus in GTK 4.0. We’d probably prefer that.
> 
> Ciao,
>  Emmanuele.
> 
> > Any insight is appreciated!
> > 
> > Warm regards
> > Philipp
> > 
> > 
> > [1]: https://github.com/p-e-w/plotinus
> > [2]: https://github.com/p-e-w/plotinus/issues/35
> > [3]: https://github.com/GNOME/gtk/commit/39d1537211501a8603f93a3196
> > b910
> > dce40e1617
> > 
> > _______________________________________________
> > gtk-devel-list mailing list
> > gtk-devel-list@xxxxxxxxx
> > https://mail.gnome.org/mailman/listinfo/gtk-devel-list
> 
> -- 
> https://www.bassi.io
> [@] ebassi [@gmail.com]
_______________________________________________
gtk-devel-list mailing list
gtk-devel-list@xxxxxxxxx
https://mail.gnome.org/mailman/listinfo/gtk-devel-list