I would actually suspect that my disappointment with sub-pixel rendering in MacOS X might be in part related to the MacBook display. (Think about it: if you need to dither away the neighboring pixels to get the anti-aliasing up to spec, you certainly lose something. Getting that right was always a thin balance.)
To be fair, I noticed the same thing on a cheapo new Inspiron used by a friend (both are 1280x800 displays, for that matter). I would be really disappointed if my next 1920x1200 machine turns out in any way like it.
If, on the other hand, you go to Las Vegas and see a hundred people playing the machines there, with noone winning anything large, you now have a new posterior probability distribution compared to the case when you just see a single gambling machine for the first time in your life, gullibly thinking you might have a chance to make a fortune. In your example of the dice, we already know the probability. In my example, and the example of life in the universe, we don't, and that's the point.
Likewise, if you roll a six-sided die a thousand times, and you don't get six any single time, you should seriously start to reconsider the hypotheses that any outcome is just as likely. You have no reason to do so with a single observation, though. It's all about Bayes, and common sense.
W/hr? Ouch, mismatching units on Mars, this can't possibly be good... I'm just waiting for "CMOS battery accidentally drained, no boot on Opportunity".
You provide the wrong link, this is the proper one. 110 W, for the complete machine, and that's AC (so even if the CPU was the only DC component in the machine, it would end up consuming more from the mains).
For comparison, I think the Core Duo TDP in that machine is something like 30 W, maybe a bit more.
n-gram based language models are nothing new. Statistics is all fun and dandy, but it's no panacea. It might just be enough to throw in an even larger corpus (something like the complete Google index), but it's still hard.
(BTW, n-gram Markov chains more or less originated in speech recognition, to get the individual phonemes right, and I'm quite sure they're doing at least something like it at the word level these days. It still sucks, as the quality users demand for proper dictation is extremely high.)
As long as the TV set supports the bandwidth and codec of the original bitstream of the video source, and you don't want any on-screen display or anything, that's OK. I think that covers the problems as well. In the analog world, the SD signal in coax was actually at least a bit similar to the "actual" video. It's a much larger step to go from an arbitrary frame buffer to a properly encoded digital signal.
Big Bang was no explosion, it was the expansion of space. The shape of space is a question that's been open to some discussion, but you should not assume that the light got away and is sitting on the "edge" somewhere (or expanding the edge), because there is no such edge. Also, during much of the initial period of the universe's existence, it was opaque -- the energy levels of matter were high enough that just about any EM radiation was continuously absorbed and re-emitted, giving us the background radiation.
The most important aspect here might also be the fact that space expansion is a local event. On a large enough "distance", the speed of that event, if we just tried to add together the relative expansion per unit length, would exceed c. It can certainly approach it. There is/should be matter much farther away than the 2 * 15 bly "bubble" that would be the theoretical maximum of matter simply going in all directions at the point of Big Bang.
I would be very surprised if Apple did NOT keep a good look on whatever issues that would cause problems for current MacOS X x86 on AMD hardware. This is not like the Apple switch from IBM to Intel, it's like Motorola PowerPC to IBM PowerPC. From this perspective, the x64 port in Leopard is far more significant, and even that transition is stated to be quite smooth.
Even NTFS does it, to some degree (with the side effect that some 3rd party/open source implementations have special problems when it comes to accessing or writing such files). As a sibling post mentions, it's not only a matter of small files, though, it's also a matter of large files that are not a multiple of a suitable power of 2 in size.
That's true for C++ (in most implementations and unless you are conscious about it, like STL, which casts to void* internally in several places), but for C#, a conversion is made back for all "reference types". That is, all non-value types basically share one set of IL and JITed code in the workset. The binary representation in the file will always be a single IL copy for List and List.
I don't see any reason to consider the Java approach superior. The backwards compatibility is only an issue if you have made (somewhat) bad choices about argument types: you generally shouldn't require an ArrayList in the first place, but just a List. Back to.NET: all the generic collections (for example) implement the corresponding non-generic interfaces, so if you used an IList in some piece of 1.1 code, you can plug in your generic List just fine.
No graphics acceleration required? The guy mentioned video, and it's bad enough trying to get it over the network, but a stupid enough framebuffer driver would be the nail in the coffin.
There be whales here! (if the shipboard computer can be tricked into allowing it, maybe we could put it in some kind of sleep state until 2061 instead...)
The knowledge still has to spread. If all developers wanted two screens, and it was just a matter of convincing management as well, then this would be more plausible. As things stand now, a lot of developers would argue that a.4 GHz increase or a 10k RPM drive or (another) 2 GB of RAM, or even a slightly larger widescreen, instead of a sensible dualmon config, would help their productivity or whatever more.
(And, yeah, I would probably get a 30" screen over my two 1600x1200, but the two 1600x1200 were far cheaper at the time of purchase, and that config would still win today.)
True, but practically we can think about something that lets 0.001 % or so of the light through. In daylight, that should be enough for the observer inside to still see what's going on (the range of usable eyesight is quite astounding), while it might be limited enough to still be hard to detect for those on the outside. (Hint, though: don't use a flashlight while you're wearing the cloak in a very dark room.)
Is this the XP where slipstreaming with.sif editing (or a smart 3rd party tool do it all) is the only realistic way to install some RAID drivers? You know, like that obscure manufacturer nVidia, with logo and all? It doesn't work for XP without tinkering. That's not the only thing. If we really look at what the specs should require, like proper hibernation, it's obvious that there are loads of supposedly "fine" XP hardware + drivers that don't really make it, and in that area XP (and drivers) have really matured since the fun days of XP vanilla (bitmap cache for GUI trashed at every hibernation with my ATI mobile chipset, the 3com bluetooth card would bluescreen on every restart from standby/hibernate, both of course logoed).
The analysis you link to does not mention the kernel. It's true that some GDI is in kernel land, but a surprising amount of resource access, like this, is not. The exploit, in its current form, is firmly in the userland part, and constrained by the security tokens of the thread and process. That's often bad enough, though.
If it does affect calc.exe, it rather seems like you have some DLL injection (keylogger/spyware, or something legit) that then causes this. If they messed up the base address, or just increased the size over a previously valid boundary, all kinds of DLLs with preferred addresses in the same region could start causing interference.
You simply have to be careful with the address space if you are a library that will be dynamically loaded in plenty of images, especially if you are loaded very early on.
(Heh, last summer, I got the genius idea that the base addresses were probably not optimal after all hotfixes and 3rd party software, so I started a gigantic rebase on the complete system32. That's a baaaad idea. I should at least have had enough sense to exclude NTOSKRNL, but I obviously didn't. Repair was fun...)
For chimps, it's rather simple. Even if every single chimpanzee on Earth died this year, it would not even be a dent in the overall statistics for the number of persons (whatever the definition) on the planet.
I agree with the sentiment, though. Extending the rights to organisms that can never really claim them themselves, is quite peculiar. On the other hand, making that the single basis could lead down the road of revoking the rights for some human beings, but I find the species barrier convenient, and it certainly currently avoids issues taking up time in courts determining who is a person and who isn't (ooops...).
It's the pigeonhole principle. If we are basically right about what combinations are possible we don't know that some exact copy of ourselves will exist anywhere. What we do know is that we will have areas of space, similar in size to our observable universe, that are completely alike. It may be the really dull areas where there is no single planet to be found (but then assumptions about a basically homogeneous universe would break down). It would be kind of cool to "know" about a copy of me, but I still find it a staggering thought that there would be copies at all. Just think about the hashing problem when you try to put them in the database of Google 'verse!
You assume a planet that's geologically active. Mars is dead. There are no great reserves of inner waste heat that are slowly leaking out. (Unless the radioactive decay alone, through conduction, would be enough to keep the temperature at that level, although I doubt it. Also remember that Mars is far smaller, decay would basically be per unit of volume, but the resulting heat is then distributed per unit of surface, over the thin "shell" that we are still discussing here.)
Ok, but then you don't only ask for protected memory, but a microkernel and lots of server processes. Changing page tables on the fly to do this, while keeping the number of processes low, is completely unthinkable on current architectures. As we have no actual production OS even close to the granularity you're requesting here, the question is not what part of protected memory MS doesn't understand. In this case, they understand, and use it, in pretty much the same way as "everyone" else. (If it had actually happened down in win32k.sys, the story would have been different.)
Ok, replying to self. The MS advisory seems to claim that IE protected mode means that it can't exploited (just crashing IE). I would doubt that this is totally true, but it's clear that exploiting it to get general access to the user's account would need some extra work.
Slashdot Mirror
What about "-site:"?
To be fair, I noticed the same thing on a cheapo new Inspiron used by a friend (both are 1280x800 displays, for that matter). I would be really disappointed if my next 1920x1200 machine turns out in any way like it.
Likewise, if you roll a six-sided die a thousand times, and you don't get six any single time, you should seriously start to reconsider the hypotheses that any outcome is just as likely. You have no reason to do so with a single observation, though. It's all about Bayes, and common sense.
W/hr? Ouch, mismatching units on Mars, this can't possibly be good... I'm just waiting for "CMOS battery accidentally drained, no boot on Opportunity".
For comparison, I think the Core Duo TDP in that machine is something like 30 W, maybe a bit more.
n-gram based language models are nothing new. Statistics is all fun and dandy, but it's no panacea. It might just be enough to throw in an even larger corpus (something like the complete Google index), but it's still hard. (BTW, n-gram Markov chains more or less originated in speech recognition, to get the individual phonemes right, and I'm quite sure they're doing at least something like it at the word level these days. It still sucks, as the quality users demand for proper dictation is extremely high.)
As long as the TV set supports the bandwidth and codec of the original bitstream of the video source, and you don't want any on-screen display or anything, that's OK. I think that covers the problems as well. In the analog world, the SD signal in coax was actually at least a bit similar to the "actual" video. It's a much larger step to go from an arbitrary frame buffer to a properly encoded digital signal.
The most important aspect here might also be the fact that space expansion is a local event. On a large enough "distance", the speed of that event, if we just tried to add together the relative expansion per unit length, would exceed c. It can certainly approach it. There is/should be matter much farther away than the 2 * 15 bly "bubble" that would be the theoretical maximum of matter simply going in all directions at the point of Big Bang.
It surely sounds like a Googely way to do it...
I would be very surprised if Apple did NOT keep a good look on whatever issues that would cause problems for current MacOS X x86 on AMD hardware. This is not like the Apple switch from IBM to Intel, it's like Motorola PowerPC to IBM PowerPC. From this perspective, the x64 port in Leopard is far more significant, and even that transition is stated to be quite smooth.
Even NTFS does it, to some degree (with the side effect that some 3rd party/open source implementations have special problems when it comes to accessing or writing such files). As a sibling post mentions, it's not only a matter of small files, though, it's also a matter of large files that are not a multiple of a suitable power of 2 in size.
I don't see any reason to consider the Java approach superior. The backwards compatibility is only an issue if you have made (somewhat) bad choices about argument types: you generally shouldn't require an ArrayList in the first place, but just a List. Back to .NET: all the generic collections (for example) implement the corresponding non-generic interfaces, so if you used an IList in some piece of 1.1 code, you can plug in your generic List just fine.
WMP only plays DVDs if you have a codec in XP. (Some) Vista (editions) include a MPEG-2 decoder. More IP licensing goodness!
No graphics acceleration required? The guy mentioned video, and it's bad enough trying to get it over the network, but a stupid enough framebuffer driver would be the nail in the coffin.
There be whales here! (if the shipboard computer can be tricked into allowing it, maybe we could put it in some kind of sleep state until 2061 instead...)
(And, yeah, I would probably get a 30" screen over my two 1600x1200, but the two 1600x1200 were far cheaper at the time of purchase, and that config would still win today.)
True, but practically we can think about something that lets 0.001 % or so of the light through. In daylight, that should be enough for the observer inside to still see what's going on (the range of usable eyesight is quite astounding), while it might be limited enough to still be hard to detect for those on the outside. (Hint, though: don't use a flashlight while you're wearing the cloak in a very dark room.)
Is this the XP where slipstreaming with .sif editing (or a smart 3rd party tool do it all) is the only realistic way to install some RAID drivers? You know, like that obscure manufacturer nVidia, with logo and all? It doesn't work for XP without tinkering. That's not the only thing. If we really look at what the specs should require, like proper hibernation, it's obvious that there are loads of supposedly "fine" XP hardware + drivers that don't really make it, and in that area XP (and drivers) have really matured since the fun days of XP vanilla (bitmap cache for GUI trashed at every hibernation with my ATI mobile chipset, the 3com bluetooth card would bluescreen on every restart from standby/hibernate, both of course logoed).
The analysis you link to does not mention the kernel. It's true that some GDI is in kernel land, but a surprising amount of resource access, like this, is not. The exploit, in its current form, is firmly in the userland part, and constrained by the security tokens of the thread and process. That's often bad enough, though.
If it does affect calc.exe, it rather seems like you have some DLL injection (keylogger/spyware, or something legit) that then causes this. If they messed up the base address, or just increased the size over a previously valid boundary, all kinds of DLLs with preferred addresses in the same region could start causing interference.
You simply have to be careful with the address space if you are a library that will be dynamically loaded in plenty of images, especially if you are loaded very early on.
(Heh, last summer, I got the genius idea that the base addresses were probably not optimal after all hotfixes and 3rd party software, so I started a gigantic rebase on the complete system32. That's a baaaad idea. I should at least have had enough sense to exclude NTOSKRNL, but I obviously didn't. Repair was fun...)
I agree with the sentiment, though. Extending the rights to organisms that can never really claim them themselves, is quite peculiar. On the other hand, making that the single basis could lead down the road of revoking the rights for some human beings, but I find the species barrier convenient, and it certainly currently avoids issues taking up time in courts determining who is a person and who isn't (ooops...).
It's the pigeonhole principle. If we are basically right about what combinations are possible we don't know that some exact copy of ourselves will exist anywhere. What we do know is that we will have areas of space, similar in size to our observable universe, that are completely alike. It may be the really dull areas where there is no single planet to be found (but then assumptions about a basically homogeneous universe would break down). It would be kind of cool to "know" about a copy of me, but I still find it a staggering thought that there would be copies at all. Just think about the hashing problem when you try to put them in the database of Google 'verse!
You assume a planet that's geologically active. Mars is dead. There are no great reserves of inner waste heat that are slowly leaking out. (Unless the radioactive decay alone, through conduction, would be enough to keep the temperature at that level, although I doubt it. Also remember that Mars is far smaller, decay would basically be per unit of volume, but the resulting heat is then distributed per unit of surface, over the thin "shell" that we are still discussing here.)
Ok, replying to self. The MS advisory seems to claim that IE protected mode means that it can't exploited (just crashing IE). I would doubt that this is totally true, but it's clear that exploiting it to get general access to the user's account would need some extra work.