🚀 go-pugleaf

On Thu, 24 Sep 1998 03:30:28 +1000, Thandor
<homer_j_simpson@my-dejanews.com> wrote: 
>Patrick <patrick@howard.genetics.utah.edu> spake thusly:
>>I'm curious about something.  I like GPL for many things but do not see 
>>it as a panacea nor as applicable to all software.  My question out of 
>>curiosity regards id Software and their non-GPL games like Quake and 
>>Quake II.  Why is noone attacking id Software for not releasing the 
>>sourcecode to their game?  Or how about attacking Applix for not 
>>releasing their sourcecode?  Same for Corel.  
>
>Because they are fundamentally different beasts. One is an API, the others
>are just applications.
>
>If QuakeII wanted to become a part of every linux desktop, then people
>would be worried. But that is not the case, so it's ok.
>
>Some people only care that the core OS is free - so they don't mind games
>like Quake, or office apps like Applix. But KDE is trying to become a core
>part of the OS, so it gets them worried.

This is a nice description of the situation. 

People may protest:
"Oh, but KDE isn't really part of the OS."
or
"Qt is just off to the side; just a library."

In one respect, they are correct.  None of this stuff (whether KDE, Qt,
or, for that matter, GNOME/Gtk) represents a piece of the operating
system kernel. 

In the same strict fashion of defining things, Linux is not "a GNU
system" (as where RMS suggests that it should be called "GNU/Linux").
(And I'd note that this *does* connect...  Wait for it...)

KDE may not be an OS, in much the same fashion that Microsoft Windows
3.1 was not an OS. 

However.

If KDE becomes a "standard" as some proponents wish to see happen, then
while KDE/Qt may never represent the kernel of an Operating System, they
could quite justly be defined as representing the "kernel" of an
"Operating Environment."

People had to write applications "for Windows 3.1." Not for the OS
(which was MS-DOS, or whatever variant thereof), but for the "Operating
Environment."  Thus, from both the points of view of applications
developers and of users, what KDE wants to be is enough *like* an
operating system as to make those that protest look like CS "purists,"
disagreeing to make themselves look important.

In effect, if KDE "succeeds," it is no longer *Linux* (or UNIX, or
POSIX, or "GNU Systems") that is the target of applications, but rather
*KDE.* And this gives TrollTech a high level of power over "the Linux
industry."  And this is what gives people cause for concern.  People
wonder why Red Hat Software isn't going along with this?  This is
surely why.

The Hurd project provides something slightly isomorphic to this (back
to GNU...); it in effect provides two "layers" of OSes:
a) There is the base level with the Mach microkernel;
b) And then the more public services provided by Hurd daemons.

The "cool thing" about Hurd is that it plans to provide all sorts of
additional OS capabilities that can be customized whilst the system is
running.  I'm more than a little skeptical that many people will use
those additional capabilities very much; that effectively provides a
"layer" that (not unlike KDE/GNOME) are "not UNIX."  For those that
want UNIX, or need to interoperate with it, the added stuff may be less
than helpful... 

And then there's GNOME.  GNOME doesn't suffer from the Qt issue, where
there's a major system component that is controlled privately.  Things
are still not fun and games; while life may be AOK from a licensing
perspective, the "higher level operating environment" gives us a
headache that just won't go away. 

The major issue: GNOME has been suffering from what might be called a
"lack of stability."

On the one hand, it's still pretty "alpha" code, and one should not act
*too* prejudicially towards code that hasn't had the time to get stable. 

However, and here comes the crux of the problem, GNOME applications have
to depend on a whole lot of "GNOME components" in the GNOME Desktop
Environment.  They need such things as ORBit, Gtk, panel, perhaps a
GNOME-aware wm, and a whole assortment of "GNOME libraries." Real Soon
Now, add "Baboon" to the list.  I'll bet that, sooner or later, some
form of OO Database will come along, and that will have to be woven in. 

This is a complex jumble (jungle?) of stuff.  And the typical GNOME
application needs the whole bunch of system components.  Which means
that they're all necessarily quite deeply interwoven.  If one of them
"breaks," the whole house of cards comes tumbling down.

One could construct an equivalent list of things for KDE. 

This all parallels the Microsoft situation where there's this single API
called "Win32" that contains and represents a large "house of cards."
It is so well constructed that it can fall down every day of the week.

I think I'd argue that the whole notion of building an "integrated
desktop environment" is something of a mistake.  Everyone (amongst some
crowds) wants "drag'n'drop" (others would call it "drag'n'drool");
they'd like One True User Interface.   

I don't think that this is a feasible project without going down to the
application level and defining things from the *application*
perspective, defining user interface as needed by applications rather
than (as actually happens) by defining the "perfect UI" and then
fiddling applications to conform with it. 

-- 
"Who is General Failure and why is he reading my hard disk ?" 
-- leitner@inf.fu-berlin.de, Felix von Leitner
cbbrowne@hex.net- <http//www.ntlug.org/~cbbrowne/lsf.html>