Re: Additional questions for the board candidates

[Federico Mena Quintero <federico ximian com>]

On Wed, 2005-11-23 at 18:56 +0100, Philip Van Hoof wrote:

> How important are desktop standards for you. How will you attempt to let
> the GNOME developers cooperate even more with the freedesktop.org
> movement? Or do you dislike that movement? In in general: What should
> GNOME "do" with fd.org?

Destkop standards are interesting when they let external developers make
assumptions.  Can I drop a file in a well-known location, and be assured
that a launcher for my app will appear in the panel/kicker/whatever?
Can I, using $foo_toolkit, cut an image to the clipboard and be assured
that $bar_toolkit will be able to paste it in an application?

Desktop standards would be *REALLY* interesting if each platform (GNOME,
KDE, Mozilla, OpenOffice.org, etc.) provided a conformance test suite.
Any of the GUI test suites based on the accessibility interfaces would
be *great* for this.  Write an LDTP/Dogtail script to cut an image in
the GIMP, paste it in OpenOffice, and ensure that it works.

Desktop standards are not interesting if they are not really implemented
by the major players, or if they are implemented with too many
idiosyncrasies or platform-specific extensions.  The MIME spec is a very
sad example of this.

Thankfully, freedesktop.org is not about de-jure standards.  What we
need is more people, in any of the platforms, to identify the issues
that would benefit from standardization, *and to go ahead and write and
implement the standards*.

GNOME should take the responsibility of figuring out what would benefit
from standardization.  It should do this by establishing close contact
with other big projects and finding common factors.  And it should help
the implementation wherever possible.  See my rant on "superheroes"
here:

http://primates.ximian.com/~federico/news-2005-06.html#20
http://primates.ximian.com/~federico/news-2005-07.html#06

In particular, the second link has a tidbit from Microsoft's infamous
Halloween Memos, which I'll quote again:

	Integrative work across modules is the biggest cost encountered
	by OSS teams.

This is still true, and it is why standardizing things is so hard - no
one wants to change their code to make use of a standard library.
Someone has to go and do it for them.  This is why Fontconfig worked.

> Second question:
> 
> What will you do to further enhance cooperation with the KDE developers?
> Will you invite them to our conferences? Will you pay their travel
> expenses? Will you let them talk on GUADEC? Will you visit their
> conferences and will you do a talk about cooperation at their
> conferences? Or will you simply disregard them and think GNOME is
> superior yadiyada (in which case I wont vote for you, by the way)?

We enhance cooperation by participating in the standards process.

And there's cross-pollination among conferences already, isn't there?

I cannot disregard them because my employer has a policy of supporting
both desktops.

> And D-BUS is moving forward rapidly. This will introduce a lot new such
> standards. Even D-BUS itself is such a standard of which it hasn't been
> said that it's "the" IPC for a typical modern GNOME application. Or is
> it ORBit-2? D-COP? I guess nobody knows.

D-bus will succeed or fail, just as ORBit/Bonobo did, depending on
whether people write actually interesting public interfaces that any
interested application can use.

Neither D-bus not ORBit are interesting by themselves.  They become
interesting when I can know that I can search my addressbook with them,
or when I can tell applications that the network went offline.

D-bus seems to be well accepted across more than the GNOME platform.  We
need to publicize the useful interfaces which applications expose
through D-bus, and see if they would benefit from standardization.

> I can imagine companies that would like to target the GNOME desktop,
> while developing solutions for their customers, would like this type of
> leadership to happen. Yet I can imagine a lot Free Software GNOME
> developers dislike "any" form of "leadership". It's not a simple problem
> to solve. Will the GNOME Foundation fill this gap? Or will the GNOME
> Foundation create a solution? How will you, provided you become board
> member, address this. Or isn't this important enough for the Board to
> discuss? Or isn't it the focus of the Board?

You mentioned Linus as being the leader for the Linux kernel.  Having a
leader works for Linux because it is constrained in function:  in
oversimplified terms, and from the viewpoint of applications, Linux only
has to provide a Unixish API that works.  Linus can then oversee whether
proposed modifications to the kernel are in line with this goal.

Would GNOME benefit from a central leader?

GNOME has become too big a platform for anyone to grasp the whole thing
in his head.  It has no bounding technical goal.

We need to keep GNOME from growing simply by accretion.

The way to grow a large system is to ensure that the different pieces
are in line with some basic principles.  That is, we have to ensure the
conceptual integrity of the whole so that the different parts match
together.  See chapter 4 of "The Mythical Man-month".

So, how do we ensure GNOME's conceptual integrity?

>From the API viewpoint, by ensuring that APIs have more or less the same
form.  From the C viewpoint, that means that objects are
reference-counted and provide accesors, in order to be friendly to
language bindings.  From the friendliness-to-developers viewpoint, by
mandating that the APIs have proper documentation and that they were
designed with an application in mind.

>From the user's viewpoint, by following through on the usability
project.  If a program is not usable, it's not fit for GNOME.

So, about a central leader - perhaps we just need to lay down our goals
clearly, write some basic principles (the GNOME Programming Guidelines
have gotten a bit stale, but they are good starting point), and guide
people so that they produce software that follows those principles and
goals.

> Fourth question (finally a non programmer question! :p):
> 
> Because I can imagine it's going to be an important project for the
> GNOME desktop and infrastructure, how will you involve yourself in the
> One Laptop Per Child concept?

I'm quite involved in the performance and memory reduction efforts.
That will help from the infrastructure side of things.

  Federico