Because of teh way that the HTML control has become the core of so many components, and because of the way the HTML control does not really know whether an object its displaying should be trusted or not, there's is a whole class of attacks that are possible on Windows that are not possible on any other platform. Even Internet Explorer on the Mac was more fundamentally secure than IE on Windows for this reason... basically, when the HTML control goes to display an object, it looks at the file or URL it came from and applies its knowledge of how applications that use the control behave (for example, it knows about the location of temp files, and mailboxes, and the Internet Explorer cache) to decide whether there was any possibility that the object might be untrusted. When this test fails, perhaps because some application puts temporary files in some unexpected place, what's known as a "cross zone" exploit can be implemented.
Every other browser defaults to treating ALL content as untrusted, and only allows shell programs and applications to *add* mechanisms to a *specific* instance of that browser. That's the only secure way to design a browser. Until Microsoft abandons the current design of the HTML control, changes the API so that applications are given control and responsibility for trust, and breaks the existing API, Windows will always be subject to far more problems than any other OS.
I've got 20G of music on my computer. I've digitized all my CDs and about 10% of my vinyl and tape, and that's 30 years worth of accumulated music... by the time I'm finished I suspect I'll be pushing the limits of a 60G iPod. Luckily I'm happy with the "shuffle" model... I actually prefer the shuffle to the full-sized iPod because the controls aren't as finicky as the stupid click-wheel.
So, without a recognizable voice, how do you plan on communicating with the technical support person once you actually get them on the line?
I assume the technical support person is a human, and not a computer running speech recognition software. Even the best speech recognition software is still pretty incompetant at dealing with any kind of accent, compared with any normal human.
Though now that you mention it... a good many technical support people seem to have trouble passing the Turing test. I mean, it's hard to imagine an actual human responding to a sentence that ends in "... the problem was still there after I rebooted my computer" with "Did you try rebooting your computer?", so you may have something there.
Knowing that of the people who bought music online iPod owners were less likely to use illegal file-sharing is still useful information.
Also: Remember that close to 100% of the people who bought iPods are likely to have used the iTunes Music Store at least once... because you get 10 downloads bundled with the iPod.
Maybe they should just start selling iLife as a service you pay $100/year for, instead of charging $80 for iLife plus $100 for.Mac and treating iLife as an annex to.mac...
They'd lose that $80 from the retail version, but given that they're bundling iLife with all Macs anyway... how many sales is that?
If you want free ad laden email get a crappy MS hotmail account.
No thanks, I get GREAT ad-free email from Google, and I don't have to pay every year to keep it. And it works transparently with Apple's Mail.app, even through a firewall. And I can use it from any web browser, these days, anywhere.
Why should I bother with a.mac email address? What's the value to me?
Simplicity. Functionality. Relatively cheap price. What is problem?
Not a lot of functionality, at least that I can see never having used it, and from outside it looks really ugly. And you really can't test it without committing to pay, psychologically, because so many of the services are things like email that you know you're not going to want to shut down in three months.
If they made it so that the long-term commitment stuff, like the email account, was yours NO MATTER WHAT... even if you decided the rest of the service sucked, I'd be more willing to try it. But not as it stands. Even if it was free, trying it out takes time.
That's the problem. It's like the problem of selling people on the Mac in the first place, except that the value is even less obvious.
So.
Tell me why I should care. Tell me why I should spend my time even trying it out. What's the great functionality you get for the $100 a year?
Your supposition would require that no record in a WMF file could be 6385492 words long - or, more specifically, that there is a known maximum less than the maximum storeable value. As Gibson mentioned, the minimum record size is 6 words, which frees up the values 0 through 5 to be chosen as your magic key (or perhaps negative numbers if you use signed values for the record size).
Actually, since the record has to be an even number of words, any odd magic number would work, as well as any number less than 6. Also, since it's a 4 byte value, half the negative integers would work as well (IIRC, the maximum address space on any Win32 system is 3 GB, so no metafile object can be bigger than that).
So a magic number like 1337 or 17 or 12345 or -42 would work just as well.
Apple is selling two services here. A moderately high bandwidth hosting service, and a convenient consumer service. Split them out, with.Mac "lite" providing all the cheap services... email, small scale file sharing, chat... and you can pay for your gigabyte of disk and terabyte of bandwidth.
But for god's sake... clean up the ugly-ass folders first. I'd be embarassed to host files on.mac with those "carbon file sharing" folders.
What is so bad about asking a fair price for a service? why should apple give it away?
Because they'd make more money from the consumer services part of the deal?
Go ahead and charge for the high bandwidth file hosting, but the rest of that stuff is cheap to implement and more profitable when you have 100 times as many eyeballs hitting your portal.
I haven't even used the "free".mac trial, despite being on my third Mac and having Macs at the office. Why? Because I can not conceive of anything it could possibly provide that's worth $100 or even $30 a year. There are dozens of free, universally available, and frankly less ugly (my god, is that "carbon file sharing" the best they can do?) equivalents for everything it offers.
And it's not because I can't afford it (Macs ain't cheap), or because I won't pay for a service (I'm paying $40 a month for a virtual colo). It's because at this point.mac looks like an obsolete prototype for the free services I'm already using, not something worth forking out real money for.
It's silly/stupid to think that DRM is unique in this way.
Um, no.
DRM is unique in that it has to operate at the OS level, and that its entire reason is to prevent the normal operation of the computer from proceeding to completion. Even anti-virus software... which I consider a bad solution to the wrong problem... can run purely at an application level, by running the potentially infected application in a sandbox, and anti-virus software can always be turned off because it's the user who makes the decision whether to run it or not.
You can run your word processor in a sandbox, you can run your web browser in a sandbox, but you can't DRM software in a sandbox, lest you bypass the DRM by hijacking its outputs! Your word processor and web browser don't contain code to disable other applications that could be triggered accidentally by a bug, but since that's the whole point of DRM software it's far more likely to happen accidentally. And in a multi-user system there's nothing a word processor in my daughter's account can do to damage anything in mine... but if that was true for DRM software you'd just have to run your "sniffer" in another account to hijack the decoded output.
Users should not be misled into thinking that strong DRM is similar in any way to applications. It has to operate at the OS level, and it has to have more rights at that level than the user themself!
I think I was invited to take part in it... at least I was invited to take part in a similar survey. I ran into a problem. On the very question of where my music comes from.
I had the choices, if I recall correctly, of "downloading from a file sharing service", "purchasing from an online music service", "ripping my own CDs", or "copying from my friends". There might have been a couple more, but you could see the idea they had. But there wasn't any option for "downloaded from artist's own website" or "purchased directly from artist". And since a good 20-30% of my music falls into those categories, I stopped there and sent mail asking for clarification.
Why? Distributing the Flip4Mac product provides a better experience for Mac users, because now we can use Windows Media files in any Quicktime-supporting program... instead of being stuck with Microsoft's application.
Well, I meant "too computationally intensive to do an adequate job of in a JIT compiler", but... hmmm... I don't know. At the very least it's probably a lot worse than O(N). And the optimal solution may still not be very good.
Apple isn't the only one who tracks your music! Whoa! If you install the Audioscrobbler plugin and run it, it sends information about all the songs you play and lets ANYONE on the Internet see what you're listing to! OMG! That's exactly what it's supposed to do!
Apple's been ramping them up in OSX lately (all files now support an arbitrary number of streams, I believe), so it will be neat to see what happens with them.
On the contrary, Apple is no longer using resource forks for anything but legacy applications and to store non-critical metadata like file icons or keywords and indexes. Instead, an application is implemented as a directory tree, with separate files for each resource.
rosetta instantly recompiles PPC code to x86, not emulate, so the only true overhead is the compilation
JIT recompilation is a really good technique, but it's nowhere near a match for optimised native code. Particularly when you have a problem like the x86 register bottleneck to deal with.
Freeport Express on the Alpha provided really good performance translating from x86 and Sparc, but Freeport Express was converting from processors with 8 registers or 32 visible registers to one with 64 registers, so you could put all your emulated registers in real registers.
This time the code you're JIT translating is optimised for a relatively register-rich environment, and optimised with the assumption that the cost for fetching data from a register is very low. Converting that code to work well where the cost for fetching data from a register may involve moving other data out of a register first so the physical register can be freed up for another emulated register is a Hard Problem.
Emulation without JIT translation typically gives you at best 1/10th the performance of native code, and where you're emulating a processor with more registers that's very optimistic. Freeport Express was able to run x86 code on an Alpha with performance comparable to contemporary Pentium processors... but for the machine I was using that Alpha had a 66% faster clock than its Pentium peer. Rosetta has a much harder job to do, and is NOT going to run translated code at the claimed 80%, my observed 60%, or even an optimistic 30% of the speed of native code.
The 80% figure people quote is not the speed of the translated code. It's the speed of a "typical application"... one that is spending most of its time in Apple's GUI libraries, which are not running under emulation.
For the MacBook, where getting rid of the the G4 memory bottleneck gives you a 6x improvement in the Stream benchmark, you may well be looking at a genuine performance improvement for "typical applications". For the G5? Unless you're almost entirely using the applications shipped with the system, you're better off waiting.
Slashdot Mirror
Because of teh way that the HTML control has become the core of so many components, and because of the way the HTML control does not really know whether an object its displaying should be trusted or not, there's is a whole class of attacks that are possible on Windows that are not possible on any other platform. Even Internet Explorer on the Mac was more fundamentally secure than IE on Windows for this reason... basically, when the HTML control goes to display an object, it looks at the file or URL it came from and applies its knowledge of how applications that use the control behave (for example, it knows about the location of temp files, and mailboxes, and the Internet Explorer cache) to decide whether there was any possibility that the object might be untrusted. When this test fails, perhaps because some application puts temporary files in some unexpected place, what's known as a "cross zone" exploit can be implemented.
Every other browser defaults to treating ALL content as untrusted, and only allows shell programs and applications to *add* mechanisms to a *specific* instance of that browser. That's the only secure way to design a browser. Until Microsoft abandons the current design of the HTML control, changes the API so that applications are given control and responsibility for trust, and breaks the existing API, Windows will always be subject to far more problems than any other OS.
I've got 20G of music on my computer. I've digitized all my CDs and about 10% of my vinyl and tape, and that's 30 years worth of accumulated music... by the time I'm finished I suspect I'll be pushing the limits of a 60G iPod. Luckily I'm happy with the "shuffle" model... I actually prefer the shuffle to the full-sized iPod because the controls aren't as finicky as the stupid click-wheel.
So, without a recognizable voice, how do you plan on communicating with the technical support person once you actually get them on the line?
I assume the technical support person is a human, and not a computer running speech recognition software. Even the best speech recognition software is still pretty incompetant at dealing with any kind of accent, compared with any normal human.
Though now that you mention it... a good many technical support people seem to have trouble passing the Turing test. I mean, it's hard to imagine an actual human responding to a sentence that ends in "... the problem was still there after I rebooted my computer" with "Did you try rebooting your computer?", so you may have something there.
beyond this, they have a very functional voice recognition system that is way faster than dialing in a million numbers
Maybe for you, this is functional. Unfortunately, while everyone's kleypad generates the same DTMF tones, not everyone's voice is as recognisable.
For me, I'd much rather dial "1" than repeat an answer because it didn't understand me the first time.
Since the resale value of an iTMS download is about the same as the resale value of a bilabial fricative, um, no...
Knowing that of the people who bought music online iPod owners were less likely to use illegal file-sharing is still useful information.
Also: Remember that close to 100% of the people who bought iPods are likely to have used the iTunes Music Store at least once... because you get 10 downloads bundled with the iPod.
And my point is that if google can provide that free mail address without making it an "ad-laden system", so can Apple.
Where the division should be, I don't know. But the prospect of coming to depend on Yet Another paid service is daunting.
Maybe they should just start selling iLife as a service you pay $100/year for, instead of charging $80 for iLife plus $100 for .Mac and treating iLife as an annex to .mac...
They'd lose that $80 from the retail version, but given that they're bundling iLife with all Macs anyway... how many sales is that?
If you want free ad laden email get a crappy MS hotmail account.
.mac email address? What's the value to me?
No thanks, I get GREAT ad-free email from Google, and I don't have to pay every year to keep it. And it works transparently with Apple's Mail.app, even through a firewall. And I can use it from any web browser, these days, anywhere.
Why should I bother with a
Simplicity. Functionality. Relatively cheap price. What is problem?
Not a lot of functionality, at least that I can see never having used it, and from outside it looks really ugly. And you really can't test it without committing to pay, psychologically, because so many of the services are things like email that you know you're not going to want to shut down in three months.
If they made it so that the long-term commitment stuff, like the email account, was yours NO MATTER WHAT... even if you decided the rest of the service sucked, I'd be more willing to try it. But not as it stands. Even if it was free, trying it out takes time.
That's the problem. It's like the problem of selling people on the Mac in the first place, except that the value is even less obvious.
So.
Tell me why I should care. Tell me why I should spend my time even trying it out. What's the great functionality you get for the $100 a year?
Your supposition would require that no record in a WMF file could be 6385492 words long - or, more specifically, that there is a known maximum less than the maximum storeable value. As Gibson mentioned, the minimum record size is 6 words, which frees up the values 0 through 5 to be chosen as your magic key (or perhaps negative numbers if you use signed values for the record size).
Actually, since the record has to be an even number of words, any odd magic number would work, as well as any number less than 6. Also, since it's a 4 byte value, half the negative integers would work as well (IIRC, the maximum address space on any Win32 system is 3 GB, so no metafile object can be bigger than that).
So a magic number like 1337 or 17 or 12345 or -42 would work just as well.
Apple is selling two services here. A moderately high bandwidth hosting service, and a convenient consumer service. Split them out, with .Mac "lite" providing all the cheap services... email, small scale file sharing, chat... and you can pay for your gigabyte of disk and terabyte of bandwidth.
.mac with those "carbon file sharing" folders.
But for god's sake... clean up the ugly-ass folders first. I'd be embarassed to host files on
What is so bad about asking a fair price for a service? why should apple give it away?
Because they'd make more money from the consumer services part of the deal?
Go ahead and charge for the high bandwidth file hosting, but the rest of that stuff is cheap to implement and more profitable when you have 100 times as many eyeballs hitting your portal.
I haven't even used the "free" .mac trial, despite being on my third Mac and having Macs at the office. Why? Because I can not conceive of anything it could possibly provide that's worth $100 or even $30 a year. There are dozens of free, universally available, and frankly less ugly (my god, is that "carbon file sharing" the best they can do?) equivalents for everything it offers.
.mac looks like an obsolete prototype for the free services I'm already using, not something worth forking out real money for.
And it's not because I can't afford it (Macs ain't cheap), or because I won't pay for a service (I'm paying $40 a month for a virtual colo). It's because at this point
It's silly/stupid to think that DRM is unique in this way.
Um, no.
DRM is unique in that it has to operate at the OS level, and that its entire reason is to prevent the normal operation of the computer from proceeding to completion. Even anti-virus software... which I consider a bad solution to the wrong problem... can run purely at an application level, by running the potentially infected application in a sandbox, and anti-virus software can always be turned off because it's the user who makes the decision whether to run it or not.
You can run your word processor in a sandbox, you can run your web browser in a sandbox, but you can't DRM software in a sandbox, lest you bypass the DRM by hijacking its outputs! Your word processor and web browser don't contain code to disable other applications that could be triggered accidentally by a bug, but since that's the whole point of DRM software it's far more likely to happen accidentally. And in a multi-user system there's nothing a word processor in my daughter's account can do to damage anything in mine... but if that was true for DRM software you'd just have to run your "sniffer" in another account to hijack the decoded output.
Users should not be misled into thinking that strong DRM is similar in any way to applications. It has to operate at the OS level, and it has to have more rights at that level than the user themself!
I've seen 512M shuffles like mine for as little as $70.
I think I was invited to take part in it... at least I was invited to take part in a similar survey. I ran into a problem. On the very question of where my music comes from.
I had the choices, if I recall correctly, of "downloading from a file sharing service", "purchasing from an online music service", "ripping my own CDs", or "copying from my friends". There might have been a couple more, but you could see the idea they had. But there wasn't any option for "downloaded from artist's own website" or "purchased directly from artist". And since a good 20-30% of my music falls into those categories, I stopped there and sent mail asking for clarification.
No response. Survey form timed out. End of story.
For a price it appears.
Err, the codecs themselves are free (as in beer) from Microsoft.
if you right-click on a .wmv file, quicktime player doesn't pull up as a valid option to open the file.
Right Click, Get Info, change the player to "WMV Player", click "Change All".
Why? Distributing the Flip4Mac product provides a better experience for Mac users, because now we can use Windows Media files in any Quicktime-supporting program... instead of being stuck with Microsoft's application.
We have XPostFacto for installing OS X on older unsupported Macs.
:)
It's time for WinPostFacto for installing Windows on newer unsupported Macs.
Well, I meant "too computationally intensive to do an adequate job of in a JIT compiler", but ... hmmm ... I don't know. At the very least it's probably a lot worse than O(N). And the optimal solution may still not be very good.
Apple isn't the only one who tracks your music! Whoa! If you install the Audioscrobbler plugin and run it, it sends information about all the songs you play and lets ANYONE on the Internet see what you're listing to! OMG! That's exactly what it's supposed to do!
Malware. Bah. People are really reaching.
Apple's been ramping them up in OSX lately (all files now support an arbitrary number of streams, I believe), so it will be neat to see what happens with them.
On the contrary, Apple is no longer using resource forks for anything but legacy applications and to store non-critical metadata like file icons or keywords and indexes. Instead, an application is implemented as a directory tree, with separate files for each resource.
rosetta instantly recompiles PPC code to x86, not emulate, so the only true overhead is the compilation
JIT recompilation is a really good technique, but it's nowhere near a match for optimised native code. Particularly when you have a problem like the x86 register bottleneck to deal with.
Freeport Express on the Alpha provided really good performance translating from x86 and Sparc, but Freeport Express was converting from processors with 8 registers or 32 visible registers to one with 64 registers, so you could put all your emulated registers in real registers.
This time the code you're JIT translating is optimised for a relatively register-rich environment, and optimised with the assumption that the cost for fetching data from a register is very low. Converting that code to work well where the cost for fetching data from a register may involve moving other data out of a register first so the physical register can be freed up for another emulated register is a Hard Problem.
Emulation without JIT translation typically gives you at best 1/10th the performance of native code, and where you're emulating a processor with more registers that's very optimistic. Freeport Express was able to run x86 code on an Alpha with performance comparable to contemporary Pentium processors... but for the machine I was using that Alpha had a 66% faster clock than its Pentium peer. Rosetta has a much harder job to do, and is NOT going to run translated code at the claimed 80%, my observed 60%, or even an optimistic 30% of the speed of native code.
The 80% figure people quote is not the speed of the translated code. It's the speed of a "typical application"... one that is spending most of its time in Apple's GUI libraries, which are not running under emulation.
For the MacBook, where getting rid of the the G4 memory bottleneck gives you a 6x improvement in the Stream benchmark, you may well be looking at a genuine performance improvement for "typical applications". For the G5? Unless you're almost entirely using the applications shipped with the system, you're better off waiting.