OT: Recommended Linux Laptops, suppliers?

[Cliff Wells]

Hi!  Sorry, I went home.  Luckily the thread didn't die in my absence
<wink>. It's a bit cold, but salvageable.  I'll try to be more vigilant.

On Tue, 2003-03-18 at 18:16, Brian Quinlan wrote:
> > No doubt they have some nice features.  They also lack some I consider
> > essential.  I don't think we're going to reach an agreement on what a
> > good price/performance ratio is.
> 
> Using Skip's numbers, the 1GHz Powerbook scores about 10% less pystones.
> I think that 10% less performance, for a similar price, is reasonable.

Agreed.  10% is rarely noticed outside of benchmarks.  As I mentioned in
a different post, I'll concede, given the requirements (battery life,
weight) that Powerbooks are cost-competitive with x86 laptops.  Given
different requirements, the picture changes quite a bit, but in this
area Apple does well.

> > Perhaps you get a bigger paycheck or
> > are happy with the same PC for longer than I am.  I spend around $400
> a
> > year to upgrade my PC (new mobo, CPU and RAM, occasionally an
> additional
> > amount for a larger HDD) and easily double my performance every year.
> > With a Mac I'd have to shell out $2000/year to maintain the same
> > situation (and given their glacial processor improvements *if* it
> could
> > be done remains questionable).  To me that's unacceptable in today's
> > commodity PC market.
> 
> We were talking about laptops here. What PC laptops have upgradeable
> motherboards?

Apparently I tried to bring Apple systems in general into it as a way to
bolster my argument ;)  Sorry, I got carried away and forayed even
further OT.  As an aside, there *was* a company near here (Everex, for
those with memories) that did laptop upgrades (yes, mobo and all), but
that's rare and I think they've gone missing anyway.

> > > Are your fonts antialiased or is every graphic object antialiased?
> And I
> > > am talking about client graphics, not window manager constructs.
> > 
> > Everything.  The fonts are AA by X (Xft extensions), but the desktop
> is
> > AA by the video driver (nVidia), which, IMHO, is where it ought to be
> > done (most modern video cards support fullscreen AA in hardware).  I'd
> > be surprised if Apple doesn't leverage this same hardware support.
> 
> Cliff, I don't think that you understand how the X architecture works.
> Here is a diagram: http://www.x.org/X11_protocol.html

And I don't think you understand what I said ;)  I'd be a fool to claim
I understand X, but that the information in that link I think I can lay
claim to <wink>.  But I think you misunderstood me (and perhaps that's
my fault).  See 2 paragraphs below.

> If you look at the protocol, there is no way for a client to specify
> that it wants to draw a line terminating at subpixel locations. This is
> actually true of all the X functions that I am aware of. The server can
> chose to render the line using antialiasing, if it wishes, but precision
> is lost between client and server (because the protocol specifies
> integers for coordinates). There is also no way to assign blending
> values when drawing. 

I think it would have been easier to have fixed this than to have
started from scratch.  It would have been a nice contribution back to
the community as well, considering what Apple has gotten from them. 
Granted, such extensions would have been incompatible, but all
extensions are (for a while) and a working implementation would have
gone a long way toward creating a new standard.  If everyone chose to
ignore those extensions, that would be their loss, Apple would be no
more incompatible than they are now and they would at least have tried.
This is what I find disappointing.

> Your point about your desktop being antialiased is irrelevant because
> your desktop is a window manager artifact and your window manager is not
> an X client.

This is what you misunderstand.  When I said "desktop" I was referring
to the area of video memory visible onscreen, *not* the window manager. 
I'll try to avoid ambiguous terminology, but it can be difficult in this
area.

Anyway, my point was that X knows *nothing* about the AA.  It's done in
hardware, at the driver level.  I don't doubt that this isn't as
accurate as being able to specify sub-pixel coordinates when drawing,
but I don't think it marks the end of high-quality graphics under X.

> Where the pixel calculations are actually done is irrelevant to the X
> (and Quartz) client. 

Not seeing your point here.

> > > There are OpenGL extensions for X windows. Why do you think that is?
> > > Because it would be way too expensive to convert OpenGL constructs
> into
> > > X primitives and send those across the network.
> > 
> > But at least they *can* be sent across the network <wink>. 
> 
> Well, they can't because no one is insane enough to do this. But you
> *could*, I guess. Just like you could just send the entire window buffer
> over in Cocoa. Doing that with a 1024x768 window @ 60fps would cost you
> 4MB/s of network bandwidth. And you'd lose a bunch of information (like
> color correction data) and optimization opportunities.

You're agreeing with me (sort of).  X doesn't do this, most other
"remote display" technologies do (although most of them use some sort of
optimization to avoid sending entire screens when only a few pixels
change).  This is why I argue that X is still the best for remote
display.

> > I've done
> > little research and it appears that Apple's remote desktop is very
> > similar to VNC or PCAnywhere.  
> 
> Apple doesn't have a remote desktop product, do they? They just have
> hooks to allow others to develop them. 

Apparently they do:

http://www.apple.com/remotedesktop/faq.html

I found it by typing "apple remote display" into Google.  /F would have
been proud.  However, as I mentioned earlier, it seems to use the same
sort of system as VNC (come to think of it, it probably *is* VNC,
relabeled by Apple <wink>).  Basically the local PC and the remote PC
share a common desktop.  This is handy, but not nearly as flexible as
being able to run a single app remotely without affecting the display of
the PC the app is running on.

> > Undoubtedly Apple figures this type of thing isn't
> > too interesting to their target market (and perhaps they'll add it
> > later, if their architecture allows) but to me it's an essential
> aspect
> > of my computing environment.
> 
> The architecture allows it and any interested party is free to implement
> it. 

X already does it and no one needs to implement anything other than
hardware drivers.

> > I have 4 PC's here at work, one in another
> > room, and I routinely run monitor apps, editors and whatnot remotely.
> 
> I do this too. I just use ssh, which I'll admit is less cool.

Actually I use ssh too.  It sets up the display and authorization
settings automatically so I can just do 'ssh user at host' followed by
'xmms' and start playing music somewhere that I can't hear it ;)  And of
course, being cool is why we do stuff like that <wink>


-- 
Cliff Wells, Software Engineer
Logiplex Corporation (www.logiplex.net)
(503) 978-6726 x308  (800) 735-0555 x308