Jef Poskanzer is a lot cooler than Wile E. Coyote, too. I've been using his tiny HTTP server for years. It does what it's supposed to, it just works, and I've never had to dig it and its rocket sled out of a mountainside.
I have had the same address since 1989, long before there WAS a spam problem. My email address was all over Usenet when Cantor and Seigel sent out their first spame, which means it's all over Google Groups. The horse is so far out of the barn its grandchildren are headed for the glue factory.
In 2000, the last time I added it all up, I was getting 300M a month *after* applying blacklists. At this point my mailserver is blocking several countries and ISPs, using multiple blacklists, and running some custom greylist software I wrote myself (for qmail... sorry, Jef), and my local mail client's only seeing 20-30 spams a day out of the hundreds of thousands (maybe as many as a million, it's too depressing to keep track) of delivery attempts that show up in my logs.
If you don't mind changing your email address now and then, more power to you, but I'm damned if I'll give the bastards the satisfaction.
A billion MIPS for defence, but not a byte for tribute!
I think you put the wrong link in there. That is a dual core CPU, but not really a G4, it's an embedded controller with a PPC core. Great for a router, not appropriate for a PowerBook.
And why, pray tell, is it not suitable for a Powerbook? Seriously? The MPC8641 has the same core as the MPC7448 (which is a G4, down to the socket, which is why it's only a slightly faster bus than the current G4), it's got ample cache, and it's got most of the interfaces that a typical PC chipset provides built in and it's low power even with those interfaces which means more power saving since you don't have to add the power consumption of those parts of the support chips. The only possible issue is that it requires a new logic board, because it's got a new pin-out, but ANY real improvement to the performance of the Powerbook will need that, whether it's a new G4, a G5, Pentium, or G3+Altivec... because that's where the bottleneck is: the slow system bus on the MPC7xxx line.
If you think a G5 Powerbook was ever a realistic possibility, then I honestly can't see how you could object to this.
What is this going to do to Linux development? Apple is the Unix that is on a roll at the moment.
Nothing. Mac OS X is a gorgeous, practical, functional, and just plain superior desktop. But (a) Linux has ironically created a certain amount of application lock-in (though much of that is due to gcc and gcc extensions, not all of it is), and (b) it's got a ways to go on the server. The two main problems are:
Performance of the standard UNIX APIs are less than stellar. In particular, Mach threads are too heavyweight.
Implementation of the standard UNIX APIs are incomplete. The one that I ran into? They don't provide a standard tape interface, so you can't run any standard UNIX streaming-tape software on a Mac. I don't know if Apple made some deal with Dantz or they just like Retrospect a whole bunch, but it sure ain't my idea of how to do backup.
Now Linux isn't my free-UNIX-of-choice, desktop or server, but while I'm using OS X as my desktop I'm not happy with Macs as servers until they do something about these problems.
The last RH version I used was RHEL v3 and, well, it's easier than Windows (that's not frigging hard) but... damn... I'm sticking to BSD.
Oh, sorry, wrong kind of troll.
As far as antivirus is concerned, that whole product category is the result of deep incompetence and corruption in the software industry. The design criteria you need to follow to handle untrusted content safely have only been around for 20 years: just follow orange-book mandatory access control standards. Don't grant an object any rights to modify non-volatile state visible from outside its classification, and don't let it request that access... only accept requests for addition privileges from an object with a more trusted classification.
That means: no ActiveX, no "open safe files", no "helper applications" that don't explicitly guarantee that they follow at least as tight a policy as the browser or mail program, no "install" buttons or links, no "print", and I'm not sure about Java or half the Javascript extensions in most browsers...
I don't know a single browser or HTML-aware mail program that actually follows this policy.
No, another heartbreaker was no G5 PowerBook, and not only that, but no POSSIBILITY of a G5 PowerBook.
That's not a heartbreak. Not unless you have unreasonable expectations... I never expected a G5 Powerbook. I can not conceive of the confusion of mind that would lead anyone to expect a G5 Powerbook.
Dual core Pentium Ms are on the roadmap. Dual Core G4s aren't.
The hell you say. Do you know something about Freescale you're not telling us?
Beyond that Steve knows a lot more about which chips have legs than you or I
I don't think Steve's pulling this decision out of a hat. I just don't believe for a minute that he's told us why he made it. Nothing he said on stage Monday justifies it, and I can't believe he's as naive about Intel's track record or the fantasy of a G5 laptop as he seemed.
So what's your money on? IBM's promising 3.2 GHz cores for Microsoft and Sony, now, and if they're not going to deliver those that's huge news. He's not going to have any intel-based Macs for the public until next June, and he hasn't actually committed to any schedule for Intel-based Powerbooks, so even if the e600 has a huge slip it's still a better bet.
No, Steve's holding something back. If that's some dark secret about IBM then you better get your wagers on the Xbox360 missing the Christmas shopping season down now.
The battery lasts a week with normal phone calls and a bit of texting.
What's the battery made of, dilithium? I used to have a phone that could go a week between charges, but it was a big old Nokia 6xxx bar with a mono display and no data service. My current phone is pretty typical in size and if I go a couple of days without charging it I'm setting myself up for a flat battery and interrupted calls.
Perhaps I make more calls than you do, but no more than I used to on the Nokia... and I don't use it for anything else.
Longhorn will still be bogged down with all the old shit and this will just be a half assed attempt to embrace, extend and exterminate other operating systems.
In the "All The World Is A Windows Machine" mentality of Microsoft, what OSes are they talking about running?
Windows.
The Win32 design pretty much precludes running multiple instances of an application inside a single Windows instance without heroic efforts. I've been using various ancestors of what became Terminal Server since NT 3.1 and it's amazing what they've had to add over the years just to get to a point where they can virtualise the desktop. And even with Terminal Server and User Switching it's a real pain to run multiple instances of a server.
That's one reason blade servers are so big: about the only practical way to stick multiple instances of IIS or any other native-Windows server in a rack space is to cram lots of computers into the space.
In UNIX, of course, you have ample solutions. UNIX apps are designed for a multiuser environment and even the most aggressively single-instance yield to the power of chroot, and if that's not good enough you have FreeBSD Jails and IBM's Penguin Farms... but for Windows, your only option is VMware and VMware performance with multiple instances of Windows sucks Chevy Tahoes through dirty syringes.
And Longhorn won't change this. Only redesigning the APIs that apps use and redesigning the apps to use the new APIs will. Either that, or an enforced isolated namespace that the app can play in by itself... and given the way Windows is such swam of codependant components that pretty much requires a hypervisor.
Their Hypervisor will enforce DRM, so even linux can't override it.
That only works if the motherboard refuses to load anything but Windows. Even Palladium didn't do that, it just said "if you don't run a 'trusted' (with chrome plated rotating finger-quotes) operating system, you don't get to read the certificates or other encryption keys in the trusted store". You're talking about something about a parsec or two beyond that: to get THAT far Microsoft would have to get all CPU and motherboard manufacturers to go along with them, and they couldn't even get that kind of support for Palladium.
Because, of course, the camera uses battery power when not in use.
It uses at least 2cm of space that could otherwise have been used for the battery. So either the phone is bigger (bad) or the battery is smaller (bad).
Features are not free, and features always seem to end up sucking battery life one way or the other.
The only heartbreak is the Steve Jobs said "3 MHz in a year" and it didn't happen. Well, hell, Pentium 4 didn't increase it's clock by 50% over the same period, either.
The 7400? It's still a killer core, the only problem with it is the slow memory bus, and that was being fixed and then some with the e600.
Apple had multiple vendors to supply their chips already. They didn't have to fixate on IBM and the 970. They chose to throw everything behind the 970, even though they knew the problems inherent in that brainiac design with its high clock and long pipelines. I can't imagine why they were even expecting IBM to produce a low power 970. That's not what the 970 is all about. That's what the G4 is about, and it does a good job of it.
So, no, they're not going to use VIA or Transmeta. Neither of them has a CPU that's anywhere near even the lowest end G4 they're using now. They will use Intel, they may use AMD, but if they're going to ignore Freescale and whinge about IBM I wouldn't bet on it.
I'd love to see Apple release OS X for non-Apple x86 hardware.
Me too. My Thinkpad isn't any cheaper than a 15" Powerbook, but it's a MUCH nicer computer... other than the fact that Windows sucks dirty swamp water through a used oil filter.
I seriously doubt we'll see them using a standard PC motherboard, chipset and BIOS
Until the G5 a Powermac was almost exactly a standard PC motherboard, and the fancy industrial design of the G5 has not proven itself quite the win they hoped. They're not going to use OpenFirmware, though their BIOS will be unlikely to such as badly as most. And they're almost certainly going to use Intel's chipsets.
I'm sure they're going to do a good job, though anyone with a Rev 1 G3 would tell you to wait for Rev 2 before jumping on the bandwagon, but they're going to be a lot more of a design, software, and system integrator company than a computer company.
Which is really a good thing, because it brings the day closer when they can go head-to-head with Microsoft as a software company.
That's what just happened, pretty much. Safari is just a shell around this code... something like Shiira could be bundled with it and you'd have the whole thing. Probably even run Shiira 1.0 on Panther that way.:)
Mac on Intel is the same thing as Mac on PPC from the perspective of open source code. Mac is still closed either way.
The first sentence is largely true, though it does open up the possibility of running commercial Linux packages under Mac OS X by porting the FreeBSD linux emulation system to it. The second sentence is at best a half-truth. Mac OS X combines open- and closed- source components, and can largely be considered a proprietary GUI and desktop running on top of open-source UNIX.
Speed is all relative to what your using the computer for, there was a recent slashdotted OS X Server test which suffered incredibly compared to Intel.
Actually the G5 held up pretty well. OS X didn't, but that's a different kettle of penguins.
Slashdot Mirror
Jef Poskanzer is a lot cooler than Wile E. Coyote, too. I've been using his tiny HTTP server for years. It does what it's supposed to, it just works, and I've never had to dig it and its rocket sled out of a mountainside.
I have had the same address since 1989, long before there WAS a spam problem. My email address was all over Usenet when Cantor and Seigel sent out their first spame, which means it's all over Google Groups. The horse is so far out of the barn its grandchildren are headed for the glue factory.
In 2000, the last time I added it all up, I was getting 300M a month *after* applying blacklists. At this point my mailserver is blocking several countries and ISPs, using multiple blacklists, and running some custom greylist software I wrote myself (for qmail... sorry, Jef), and my local mail client's only seeing 20-30 spams a day out of the hundreds of thousands (maybe as many as a million, it's too depressing to keep track) of delivery attempts that show up in my logs.
If you don't mind changing your email address now and then, more power to you, but I'm damned if I'll give the bastards the satisfaction.
A billion MIPS for defence, but not a byte for tribute!
In terms of my daily life Google is worth much more [than Time Warner]
I dunno. It's Time Warner that gets me to Google.
I think you put the wrong link in there. That is a dual core CPU, but not really a G4, it's an embedded controller with a PPC core. Great for a router, not appropriate for a PowerBook.
And why, pray tell, is it not suitable for a Powerbook? Seriously? The MPC8641 has the same core as the MPC7448 (which is a G4, down to the socket, which is why it's only a slightly faster bus than the current G4), it's got ample cache, and it's got most of the interfaces that a typical PC chipset provides built in and it's low power even with those interfaces which means more power saving since you don't have to add the power consumption of those parts of the support chips. The only possible issue is that it requires a new logic board, because it's got a new pin-out, but ANY real improvement to the performance of the Powerbook will need that, whether it's a new G4, a G5, Pentium, or G3+Altivec... because that's where the bottleneck is: the slow system bus on the MPC7xxx line.
If you think a G5 Powerbook was ever a realistic possibility, then I honestly can't see how you could object to this.
Will the JVM or Rosetta have more overhead? They're both interpreters, after all.
What is this going to do to Linux development? Apple is the Unix that is on a roll at the moment.
Nothing. Mac OS X is a gorgeous, practical, functional, and just plain superior desktop. But (a) Linux has ironically created a certain amount of application lock-in (though much of that is due to gcc and gcc extensions, not all of it is), and (b) it's got a ways to go on the server. The two main problems are:
Performance of the standard UNIX APIs are less than stellar. In particular, Mach threads are too heavyweight.
Implementation of the standard UNIX APIs are incomplete. The one that I ran into? They don't provide a standard tape interface, so you can't run any standard UNIX streaming-tape software on a Mac. I don't know if Apple made some deal with Dantz or they just like Retrospect a whole bunch, but it sure ain't my idea of how to do backup.
Now Linux isn't my free-UNIX-of-choice, desktop or server, but while I'm using OS X as my desktop I'm not happy with Macs as servers until they do something about these problems.
The last RH version I used was RHEL v3 and, well, it's easier than Windows (that's not frigging hard) but... damn... I'm sticking to BSD.
Oh, sorry, wrong kind of troll.
As far as antivirus is concerned, that whole product category is the result of deep incompetence and corruption in the software industry. The design criteria you need to follow to handle untrusted content safely have only been around for 20 years: just follow orange-book mandatory access control standards. Don't grant an object any rights to modify non-volatile state visible from outside its classification, and don't let it request that access... only accept requests for addition privileges from an object with a more trusted classification.
That means: no ActiveX, no "open safe files", no "helper applications" that don't explicitly guarantee that they follow at least as tight a policy as the browser or mail program, no "install" buttons or links, no "print", and I'm not sure about Java or half the Javascript extensions in most browsers...
I don't know a single browser or HTML-aware mail program that actually follows this policy.
No, another heartbreaker was no G5 PowerBook, and not only that, but no POSSIBILITY of a G5 PowerBook.
That's not a heartbreak. Not unless you have unreasonable expectations... I never expected a G5 Powerbook. I can not conceive of the confusion of mind that would lead anyone to expect a G5 Powerbook.
Dual core Pentium Ms are on the roadmap. Dual Core G4s aren't.
The hell you say. Do you know something about Freescale you're not telling us?
Beyond that Steve knows a lot more about which chips have legs than you or I
I don't think Steve's pulling this decision out of a hat. I just don't believe for a minute that he's told us why he made it. Nothing he said on stage Monday justifies it, and I can't believe he's as naive about Intel's track record or the fantasy of a G5 laptop as he seemed.
So what's your money on? IBM's promising 3.2 GHz cores for Microsoft and Sony, now, and if they're not going to deliver those that's huge news. He's not going to have any intel-based Macs for the public until next June, and he hasn't actually committed to any schedule for Intel-based Powerbooks, so even if the e600 has a huge slip it's still a better bet.
No, Steve's holding something back. If that's some dark secret about IBM then you better get your wagers on the Xbox360 missing the Christmas shopping season down now.
The battery lasts a week with normal phone calls and a bit of texting.
What's the battery made of, dilithium? I used to have a phone that could go a week between charges, but it was a big old Nokia 6xxx bar with a mono display and no data service. My current phone is pretty typical in size and if I go a couple of days without charging it I'm setting myself up for a flat battery and interrupted calls.
Perhaps I make more calls than you do, but no more than I used to on the Nokia... and I don't use it for anything else.
They can make it illegal not to have it.
You're claiming that Microsoft can make it illegal to run any operating system but Microsoft Windows anywhere in the world?
And people say Jobs emits a reality distortion field...
Longhorn will still be bogged down with all the old shit and this will just be a half assed attempt to embrace, extend and exterminate other operating systems.
Got it in one.
In the "All The World Is A Windows Machine" mentality of Microsoft, what OSes are they talking about running?
Windows.
The Win32 design pretty much precludes running multiple instances of an application inside a single Windows instance without heroic efforts. I've been using various ancestors of what became Terminal Server since NT 3.1 and it's amazing what they've had to add over the years just to get to a point where they can virtualise the desktop. And even with Terminal Server and User Switching it's a real pain to run multiple instances of a server.
That's one reason blade servers are so big: about the only practical way to stick multiple instances of IIS or any other native-Windows server in a rack space is to cram lots of computers into the space.
In UNIX, of course, you have ample solutions. UNIX apps are designed for a multiuser environment and even the most aggressively single-instance yield to the power of chroot, and if that's not good enough you have FreeBSD Jails and IBM's Penguin Farms... but for Windows, your only option is VMware and VMware performance with multiple instances of Windows sucks Chevy Tahoes through dirty syringes.
And Longhorn won't change this. Only redesigning the APIs that apps use and redesigning the apps to use the new APIs will. Either that, or an enforced isolated namespace that the app can play in by itself... and given the way Windows is such swam of codependant components that pretty much requires a hypervisor.
Their Hypervisor will enforce DRM, so even linux can't override it.
That only works if the motherboard refuses to load anything but Windows. Even Palladium didn't do that, it just said "if you don't run a 'trusted' (with chrome plated rotating finger-quotes) operating system, you don't get to read the certificates or other encryption keys in the trusted store". You're talking about something about a parsec or two beyond that: to get THAT far Microsoft would have to get all CPU and motherboard manufacturers to go along with them, and they couldn't even get that kind of support for Palladium.
They didn't have cameraphones back in the stone age when I had a baby. But that didn't stop me from buying a separate camera.
Because, of course, the camera uses battery power when not in use.
It uses at least 2cm of space that could otherwise have been used for the battery. So either the phone is bigger (bad) or the battery is smaller (bad).
Features are not free, and features always seem to end up sucking battery life one way or the other.
how long before the average cell phone has basic video recording capabilities?
I doubt I'll ever get a cellphone that has a camera, let alone video, built in. Cellphones already have too many battery-hungry features as it is.
Give me bluetooth devices I can use WITH my phone instead.
after the back-to-back heartbreaks...
What effing heartbreaks?
The only heartbreak is the Steve Jobs said "3 MHz in a year" and it didn't happen. Well, hell, Pentium 4 didn't increase it's clock by 50% over the same period, either.
The 7400? It's still a killer core, the only problem with it is the slow memory bus, and that was being fixed and then some with the e600.
Apple had multiple vendors to supply their chips already. They didn't have to fixate on IBM and the 970. They chose to throw everything behind the 970, even though they knew the problems inherent in that brainiac design with its high clock and long pipelines. I can't imagine why they were even expecting IBM to produce a low power 970. That's not what the 970 is all about. That's what the G4 is about, and it does a good job of it.
So, no, they're not going to use VIA or Transmeta. Neither of them has a CPU that's anywhere near even the lowest end G4 they're using now. They will use Intel, they may use AMD, but if they're going to ignore Freescale and whinge about IBM I wouldn't bet on it.
The 970 didn't scale as far as even the P4
The 970 got, what, 30% over 2 years? I don't think the P4 did that well over the same time.
nevermind the Opteron
Steve doesn't mind the Opteron, but he's not using it.
Also it allows the client company to produce faster, cheaper computers.
Maybe. I'm not convinced... the G5 actually got faster faster than the P4, even if it didn't get faster as fast as Steve predicted.
If he'd said "4 GHz Pentiums in a year" he'd have been just as POed at Intel.
I'd love to see Apple release OS X for non-Apple x86 hardware.
Me too. My Thinkpad isn't any cheaper than a 15" Powerbook, but it's a MUCH nicer computer... other than the fact that Windows sucks dirty swamp water through a used oil filter.
I seriously doubt we'll see them using a standard PC motherboard, chipset and BIOS
Until the G5 a Powermac was almost exactly a standard PC motherboard, and the fancy industrial design of the G5 has not proven itself quite the win they hoped. They're not going to use OpenFirmware, though their BIOS will be unlikely to such as badly as most. And they're almost certainly going to use Intel's chipsets.
I'm sure they're going to do a good job, though anyone with a Rev 1 G3 would tell you to wait for Rev 2 before jumping on the bandwagon, but they're going to be a lot more of a design, software, and system integrator company than a computer company.
Which is really a good thing, because it brings the day closer when they can go head-to-head with Microsoft as a software company.
That's what just happened, pretty much. Safari is just a shell around this code... something like Shiira could be bundled with it and you'd have the whole thing. Probably even run Shiira 1.0 on Panther that way. :)
Does this make it easier for adware and spyware writers at all?
It would be a good idea to do a security review of the code base.
Mac on Intel is the same thing as Mac on PPC from the perspective of open source code. Mac is still closed either way.
The first sentence is largely true, though it does open up the possibility of running commercial Linux packages under Mac OS X by porting the FreeBSD linux emulation system to it. The second sentence is at best a half-truth. Mac OS X combines open- and closed- source components, and can largely be considered a proprietary GUI and desktop running on top of open-source UNIX.
Speed is all relative to what your using the computer for, there was a recent slashdotted OS X Server test which suffered incredibly compared to Intel.
Actually the G5 held up pretty well. OS X didn't, but that's a different kettle of penguins.