LinuxWorld.

[Alan DuBoff]

Deirdre Saoirse wrote:

> You're not supposed to make those announcements until tomorrow. :)
> 
> But it IS cool.

I'm not exactly sure what we have won "best of" at, it might be the best pile
of resistors for all I know...;-) I don't think it's any big secret, CES seems
to pick that stuff a month or two in advance. So it's hard to say how cool it
is yet (I'm a skeptic I 'spose).

> What, and rewrite what they paid $300M for? Why? It's better than anything
> out there, mostly because all of it is tied to poorly thought-out X-based
> stuff. And, I might add, the NeXT interface was done before X was stable.
> If it ever got to the point of "stable."

I'm not so sure it's the best, in the sense that it leaves Apple with holding
a large piece of the pie, since there doesn't seem like much for the community
to reuse as Apple's efforts are aimed at their hardware specific. Yes, I know
they have an Intel version that is rumored to be lingering, and maybe they
will show it at MacWorld, I don't know...but I don't see an Intel version
gaining them that much market. The Intel market is aimed strongly at Linux,
and because that effort is going well nowdays, Apple could have gained a lot
by using a Linux kernel, and they would gain a great deal of SMP work at the
same time (very little SMP machines in the actual workplace for Apple
hardware).

I agree that the interface is the best, in the sense of the actual GUI, but it
still has some problems, and I'm not clear on how the OpenSource community can
benifit from Apple's work without reinventing the same wheels. What they have
gotten right is the glue layer between the GUI and filesystem. IOW, if you cd
to your Desktop directory in a terminal and perform a mkdir, a new object
appears on your Desktop of the same name, and if you rename a folder on your
Desktop, the actual directory name changes as well. IBM almost had it right
with the Workplace Shell on OS/2, but it was hacked together and it was
dangerous to manually change things in the actual filesystem, more often it
became corrupt over time and needed to be rebuilt.

However, Apple appears to be using some type of polling for the "glue layer",
and that doesn't seem very effecient, because there are noticable delays.

> Better idea: create something that's open source a la helping out the
> GnuSTEP project.

I don't have a preference to which project, but it would be nice if Apple
could help out on SOME project that the OpenSource community could benifit
from, that would only help the community out. This is exactly why I think they
should have gone with a Linux kernel, then they could let the kernel be done
by the community and both parties would benifit. It seems like the best of
both worlds.

> > As it is, OSX can't use the ports from FreeBSD or NetBSD as they
> > modified the directory structures and kernel.
> 
> Wrong order: FreeBSD and NetBSD came *later*, not earlier.

???? I might be missing your point. I was just referring to the ports packages
on FreeBSD or NetBSD in the sense of usable source code, but Apple can't use
that as the changes they have made to the filesystem and/or kernel changes are
not compatible with BSD. BTW, NetBSD certainly came later, but the ppc project
appears to be substantially ahead of the FreeBSD ppc project and supports a
large portion of the PowerMac hardware as well as the PowerComputing machines,
both lines back to the 601. I was almost going to install NetBSD last week and
was checking it out, but in the end I'm better off with Linux, IMO. My
colleagues think I'm sacrilege because I run Linux on my PowerBook, but Linux
actually works as well or better than the alternatives for me, and I have all
my familiar apps running on it.

I compiled XFree86 on Mac OSX (latest 4.02 code), and I can run X on it, but I
can't run X and the OSX GUI together (aqua I believe is the component), so
whats the point? I would rather my core was Linux so I can use all the
available Linux code. 

> New meme: Debian Darwin: because there's no reason not to.

It works for me.

> Well, I'm running Linux on x86.

Yes, all my other systems are x86 boxes, some dual and some uni processors.
But I was given the choice of another decked out Intel box, or a Mac, so I
opted for a PowerBook. I have most of my apps and code and need to install the
ppc cross compiler to target the 823 ppc chip. Then I'll be able to do most
everything I can do on my Intel box, on the PowerBook, which lasts for twice
as long on a battery as equivilant intel based laptops. It's good hardware to
run Linux on, even though it's only a 500mhz G3, it scoots along fine.

One aspect I hate about the Mac (and this could just be me), but they seem to
use a lot of ms software, and that is wrong. At least if I run Debian on my
PowerBook, there is no ms software. Even OSX only comes with IE for the
browser. Doesn't seem like Mozilla is available for OSX yet. I seem to always
come back to the same place, Linux appears to be my solution. And now that
OpenOffice is going, I can get it for ppc Linux.

-- 

Alan DuBoff
Software Orchestration, Inc.