who you calling limited?

Apple’s Newton Just Won’t Drop

the Newton’s limited processing power (160-MHz processor)

is anyone else shaking their head at the thought of 160 MHz being considered “limited”?

I’ve had every one of these from the OMP to the 2000 (which I stlll have). Apple really missed out o this. I’ll admit it had its flaws but some were overblown (the handwriting recognition was quite good on the later models). The Palm, of which I have had and disposed of 3 different models, is nothing more than a battery-backed roll of paper for notes, no intelligence whatever, other than in the mind of designer who made it shirt-pocket-sized and the marketeers who priced it within the geek gadget limit.

from Frank

i wish i had a camera

In the parking deck of Swedish Hospital, there is a set of disused gas pumps, for what reason I don’t know, but inside the chainlink compound with them is a very actively used bike rack more than half full of bikes.

The times they have a-changed . . . .

I misspake: it’s Classic, not Aqua, that’s the hog

OS X load w/o Classic

I never thought about this, since Classic is pretty unobtrusive (you never see it as an application, just as a framing environment for Classic apps). I rebooted this OS X box today, and when I logged in, I stopped the Classic environment from starting (usually I load it at login time). Now as you can see, the load average is quite low on the graph.

So an Xserve might not come with a performance penalty after all.

I thought Classic was supposed to “sleep” when it wasn’t in use.

CNN.rss exists, just not at CNN.com

Index of /~jacoby/XML

It turns out someone is generating RSS feeds for various sections of CNN.com content, as well as a host of other sites. Now, on the one hand, the folks at CNN don’t have to host or manage this stuff, but I would think they would want to know how popular or useful a feature like this is, and they can’t know that unless they run it themselves.

<update> Feeds are in development: take a look at this page.

odd: why would Zoe need Aqua/Quartz to run?

As another datapoint on OS X and Aqua resource usage, I noticed that having Zoe running consistently kept load average around 1 or slightly above. Now that I have shutdown my login session, it won’t run at all. I get these errors:

kCGErrorIllegalArgument : initCGDisplayState: cannot map display interlocks.
kCGErrorIllegalArgument : CGSNewConnection cannot get connection port
kCGErrorIllegalArgument : CGSNewConnection cannot get connection port
kCGErrorInvalidConnection : CGSGetEventPort: Invalid connection

I don’t know why I have connection failures. I didn’t think you could display an Aqua session over a network connection (like you can with X).

Google found this tantalizing clue:
This is what you typically get when you try to start a quartz gui
application while logged in via ssh. So your configure script starts a
gui application which it certainly is not supposed to do.

So Zoe is evidently calling on some elements of the Quartz UI and failing. This removes a potentially useful application for Zoe: I could envision Zoe running on a group mailserver, allowing a group of users to view their mail in Zoe instead of in the old linear clients. It seems silly to think you need to run the Quartz stuff on a potentially headless system, especially if it chews up 25% of your CPU.

Update: I discovered it was Classic, not Aqua, that was churning away. So leaving an Aqua session logged in, annoying and non-secure as that as that may be, won’t be a resource drain.

how expensive is Aqua in OS X?

Quite, at least in 10.1.5. I have been meaning to test this for awhile. I logged out of my OS X machine about 2:30 this afternoon and kept an eye on the load graph. Without Aqua running, utilization drops to UNIX-like levels (ie, 0.0).

load graph

Even with the monitor off and no one sitting at the machine, it uses 20 – 30% of the CPU just being. As it is now, I still have network services (I have an NFS volume mounted from it on my laptop), and I can quite happily run interactive processes on it.

Seems like a good argument in favor of logging out at the end of the day.

Update: I discovered it was Classic, not Aqua, that was churning away. So leaving an Aqua session logged in, annoying and non-secure as that as that may be, won’t be a resource drain.

PS: that long load spike from 4 AM to almost 6 AM is a fink update I have in root’s crontab.

source /sw/bin/init.csh; /sw/bin/fink selfupdate-cvs && /sw/bin/fink update-all -y

It will update fink itself from CVS, then update any installed packages from the latest CVS checkins.

PPS: I keep this graphs here.