Re: Underlying DE for the Fedora Workstation product

[Giovanni Campagna <scampa giovanni gmail com>]

2014-02-04 Alexander GS <alxgrtnstrngl gmail com>:
> CC'd over from the Fedora Desktop developers mailing list:
>
> On Mon, 2014-02-03 at 19:04 -0600, Michael Catanzaro wrote:
> > This is a very contentious topic, and you're promoting a minority view
> > (I suspect GNOME and KDE are much more popular in Fedora than the
> > other desktops), so lots of disagreement is to be expected.
>
> > On Mon, 2014-02-03 at 12:22 -0500, Alex GS wrote:
> > > Gnome Shell is a feature/product/package focused on mobile interaction
> > > for hybrids and touch enabled devices including tablets.
> >
> > GNOME (capitals please, same for MATE) is focused on desktop and laptop
> > computers, including laptops with touchscreens. GNOME has to support
> > touchscreens well because Windows has gone that route, and 90% of
> > laptops ship with Windows.
> >
> > Tablets are a secondary concern because very, very few people are
> > running GNOME on tablets. They're basically touchscreen laptops without
> > keyboards, though, so I don't think they're much of a stretch.
> >
> > Typically I associate the word mobile with phones, and nobody runs GNOME
> > on phones. A new startup, Endless Mobile, is trying to. I wish them
> > well, but I've yet to see reason to believe that will work well. (Which
> > is fine, since they're new.)
> >
> > Anyway, it sounds like you're spot-on part of the target audience for
> > GNOME Classic. I'm sure the developers would be interested in feedback
> > on why that environment doesn't currently meet your needs, and how it
> > might be improved to do so.
>
> If you look at MATE's road-map you'll quickly see most objections
> against it aren't valid.  They're actively moving closer to core GNOME
> infrastructure such as GTK3, systemd and Wayland as well as defaulting
> to current GNOME packages in increasing numbers.  Several GNOME packages
> from git.gnome.org have also been adopted by MATE's developers such as
> gnome-main-menu.
>
> If you look at www.ohloh.com below  MATE is a "very active" project with
> over 2 million lines of code, 64 contributors and the last commit was
> made 4 days ago.  GNOME 2 is alive and thriving and evolving with a
> large user-base among the top distributions Arch Linux, Linux Mint,
> Debian, Ubuntu 14.04, OpenSUSE and many others including Fedora. Compare
> this to GNOME Shell which has only around 81K lines of code, twice as
> many contributors at 157 and the last commit was 2 days ago.
>
> http://www.ohloh.net/p?ref=homepage&q=mate+desktop
> http://www.ohloh.net/p?query=gnome+shell&sort=relevance
>
> This presents a very complex situation for GNOME.  What does does GNOME
> do if they have two active thriving desktop products on the market
> coexisting in parallel?  Clearly GNOME Classic hasn't addressed the
> traditional desktop use-case and isn't seen as a GNOME 2 replacement.

Right, you say GNOME 3 Classic hasn't addressed your use case. But
would you care to explain how it is different from an hypothetical
GTK3 Mutter-based MATE with the GNOME 3 applications?
GNOME 3 Classic has a window list, traditional menus, static
workspaces, a window-based Alt-Tab and a huge list of addons and panel
applets (they're known as extensions and they're provided by third
parties, but they are the same thing). And really, GNOME 3 (Core or
Classic) is far more configurable than GNOME 2 has ever been: you can
move your panel, hide it, change the contents, move the contents, move
the notifications...
How is that not the "traditional desktop metaphor"? How is that
different from MATE? What do we gain by dropping all that on the
floor?
Should we readopt MATE (well, mate-panel and mate-applets, if I
understand your proposal correctly, and maybe libwnck), how would we
find the resources to maintain two different compositor
infrastructures, two very different styles of handling the desktop
shell at the technical level (one is a compositor plugin, the other is
a separate process talking EWMH)?
What about wayland: who will implement the desktop shell protocol in
gnome-panel and mutter (provided one complete enough is ever developed
by weston)? Note that for GNOME 3 we don't need it, because the shell
and the compositor are in the same process.
Not to mention all the other session services currently provided by
the Shell, such as the screensaver, the keybinder and the
screenshotter: who would reimplement these in a way compatible with
GNOME 3 Core, and keep them updated as the interfaces change?
What about the feature difference: is it acceptable not to have
integrated chat in the hypothetical GNOME 2 interface? Or removable
disk notifications? Or a screen recorder? Or even a bluetooth status
indicator?
None of that is covered in the roadmap, by the way. On the other hand,
what is covered in the roadmap is a lot of work in the low-level
components and the applications - both of which are not part of the
proposal, and I believe would be rejected straight away.

Yes, I understand there is a need from some people for a "traditional
desktop metaphor", because change after 20+ years is not easy, but we
already have that, in the form of GNOME 3 Classic. And if anything is
lacking, please file a bug.

Thanks

Giovanni