Saturday, 19 October 2002
I don't understand how I missed this Kuro5hin article on pecan pancakes. Read the comments for full effect.
11:38 AMSlava Karpenko discusses the failings of the kludgy OS X Mach-O ABI. I wonder if the move to OS X on 64-bit PowerPC will permit Apple to fix the ABI brokenness reasonably seamlessly?
Tuesday, 15 October 2002
Archipelago 2.0
Daniel Berlinger released Archipelago 2.0 today. So far, so good. It connected immediately to Radio running on the local machine, and it's extremely responsive compared to Radio's Web interface. So nice to write in a real desktop app instead of a text field inside a Web browser. I'll definitely be doing my writing in Archipelago from now on—unless something better comes along.
On OS X there are a large number of cosmetic anomalies, but the functionality works, and that's a good thing. Archipelago doesn't let you create a post with an empty title, and if your weblog doesn't use titles like mine does, that's a bad thing. If I try to remove the title, I end up with a period at the beginning of the post. Boo.
Is there no way from the MetaWeblog or Radio APIs to get a list of posts? I went through a lot of trial and error to figure out that my last weblog post was number 529.
The PowerPC 970 is happening, due in Macs by this time next year. After seeing the initial announcement, Tuesday I spent some time going through IBM's history of making RISC chips from 1990 onwards. Until the two lines were integrated with the POWER4, the high-end chips were split along lines of commercial (AS/400 or iSeries, some RS/6000 or pSeries) versus technical (RS6K/pSeries) workloads.
I've posted this a couple of other places, sorry if you've seen it before. To complete the timeline of IBM RISC chips:
1990 – POWER (32-bit POWER, multichip)
1992 – POWER (32-bit POWER)
1993 – PowerPC 601, 602? (modified 32-bit PowerPC)
1994-5? – A30/Muskie (64-bit Amazon, 64-bit PowerPC, multichip)
1994 – A10/Cobra (64-bit Amazon, 64-bit PowerPC)
1995 – POWER2 (32-bit POWER, multichip)
1995 – PowerPC 603, 604 (32-bit PowerPC)
1996 – PowerPC 603e/604e (32-bit PowerPC)
1996 – POWER2 SC (32-bit POWER)
1997 – RS64 (64-bit Amazon, 64-bit PowerPC, 32-bit PowerPC)
1998 – PowerPC 750 (32-bit PowerPC)
1998 – RS64-II (64-bit Amazon, 64-bit PowerPC, 32-bit PowerPC)
1999 – POWER3 (64-bit POWER, 32-bit POWER, 64-bit PowerPC, 32-bit PowerPC)
1999 – RS64-III (64-bit Amazon, 64-bit PowerPC, 32-bit PowerPC)
2000 – POWER3-II (64-bit POWER, 32-bit POWER, 64-bit PowerPC, 32-bit PowerPC)
2000 – RS64-IV (64-bit Amazon, 64-bit PowerPC, 32-bit PowerPC)
2001 – POWER4 (64-bit Amazon, 64-bit POWER, 32-bit POWER, 64-bit PowerPC, 32-bit PowerPC)
2002 – POWER4-II [expected]
2003 – PowerPC 970 (64-bit PowerPC, 32-bit PowerPC, AltiVec)
Wes pointed to an interesting article about the PowerPC's use in AS/400 and iSeries machines. There's some good history (more AIX-centric) here, and some more on the early AS/400 RISC processors here.
Monday, 14 October 2002
Pester 1.0 released. It's simple, easy to use, and mostly bug-free. May I never miss the bus again.
I can safely add NSTimer to the list of classes with bad documentation: the ownership model is essentially undocumented, and I spent several hours tracking down a crasher that turned out to be timer-related. Timers retain themselves while they're active, but something else (the runloop?) retains them as well. When you invalidate a timer or it expires, the retain count is 1, but if you try to release it, a bit later your program goes kaboom. The timer is autoreleased, so the best way to establish ownership is to retain it when you create and schedule it (retain count: 3), then perform an invalidate (has no effect if timer is already invalid) and release when you're done, relying on the later autorelease to perform a dealloc.
