On Wed, Apr 8, 2015 at 7:42 AM, Michael Catanzaro <mcatanzaro(a)gnome.org> wrote:
On Wed, 2015-04-08 at 09:52 +0200, Florian Weimer wrote:
> Is this really a good use of development resources? I mean,
> developers are the primary user group, is it unreasonable to expect
> that they will use a shell once or twice per year to perform the
Even if we only cared about developers, I would say that I do not
expect developers to use the terminal or to know that fedup exists.
But this is the beginning of the PRD we agreed on:
"We want to create a stable, integrated, polished and user friendly
system that can appeal to a wide general audience. The Fedora
Workstation working group will have a special focus on providing a
platform for development of various types of applications."
Developers are a "special focus" and that warrants making some
decisions that we wouldn't otherwise, e.g. I'm interested in
installing more command-line development tools like gcc by default.
But it would be a mistake to take this so far that we compromise user-
friendliness or our appeal to a "wide, general audience." That's one
of the reasons that developers are using Macs instead of Fedora, after
The #1 thing Apple does to achieve polish is say no. This is so
pervasive it's thematic. But constraining the context to install,
updates, and upgrades, both Apple and users gain a lot through simply
~ 90% of Custom partitioning is incompatible with stable, integrated,
polished, and user friendly. It's a massive amount of feature sprawl,
permitting arbitrary layouts that bind the bootloader, the user,
backup/restore software, offline updates, and fedup upgrades, into
understanding and supporting. There are nearly endless bugs in this
Much of what custom permits, binds us to one of two choices:
supporting arbitrary layouts, or broken functionality as a result of
them. And I've lost count how many of these bugs I've filed, let alone
how many I keep running into and just don't have the will power to
keep filing because so many keep getting marked as not bugs, or not a
priority to ever fix.
Windows and Android do stateless better. Apple does updates, backup
and restore better. Their approaches are completely different in this
area, except with one respect: they both have a lot of discipline to
just say no. So there's room for Fedora to be different and better
than OS X or Windows in this area, but not until we create a spec or
standard and stop giving users razor blades and telling them to go
play on the freeway.
And that spec/standard's mandate would be looking at a dozen packages,
how they all affect each other in terms of UI/UX and how to constrain
what's permitted up front, in order to gain stable, user friendly,
polish at the back end as a result of sane focus.