AFAIK, other than GoPro and cell phones, no cameras use micro-SD. Its a terrible, fragile form factor that is used only in tiny devices that don't have space for a real SD card slot.
That's a showstopper for me, and probably others. With Comodo, I would have to buy a wildcard for hundreds of dollars instead of a few free certs from StartSSL. TLS just went from self-evident to unaffordable and out of reach.
Any alternatives out there that are free and provide server *and* client certificates which are valid for at least 12 months (letsencrypt fanboys, don't bother)...?
No. All the other free certificates are limited to 90 days. The net effect of this decision is that only big companies and people with too much free time can afford TLS.
Not really. They design their own chips because their requirements for each device can be optimized as opposed to accepting whatever design Samsung or even Qualcomm had made to work with a vast array of different devices and customers. There's no reason they could not extend that to laptops.
Of course they could. My point is that they would end up needing to modify the cores themselves significantly to ramp up the core count, and that those sorts of changes would, I suspect, be too significant to pay off when you're talking about a laptop.
And no, the reason they design their own chips is that they can blow the doors off of what the other companies achieve in terms of power consumption by hand-optimizing the heck out of the core designs. On cell phones, that makes a big difference, and in quantities of tens of millions, the R&D cost per unit is small. On laptops, that makes a much smaller difference (because the batteries are huge by comparison) and the R&D cost per unit is relatively large.
Why would software be emulated and why would the clock speed be half?
Because nobody will have their software already recompiled for a platform that doesn't exist yet. And the clock speed would need to roughly double if you want emulated code to even approach the same per-core performance that Macs have now. They would need to increase by at least a factor of 1.5 just to get the same per-core performance for native code.
Apple has done a lot of work in the software side of OS with Grand Central Dispatch to ensure that OS and programming language knows how to use and optimize multiple cores.
And yet every time I open up my Xcode project at work, Xcode sits there with a single CPU code pegged at 100% for a couple of minutes just to load the project, and several more minutes before it stops SPODing long enough to be usable, and basically the CPU is pegged at 100% for about an hour before indexing finishes. Real-world code doesn't always parallelize easily.
As for heat issues using discrete GPUs, remember again Apple might not be using the laptop for high end gaming. They can pick a low end GPU with just enough power to handle the display.
To replace existing machines, the GPU would need to be at least as fast as what they're shipping now....
Mars gets almost half as much light as Earth (about 43%, to be more precise). Anything than can grow in partial shade should be able to grow on Mars without the need for artificial light supplementation. Fruits and root vegetables would be more problematic, of course, and would require solar concentrators, but that's still only a 2:1 ratio of ground space versus crop coverage, which isn't *that* insane.
not enough gravity.
We know the effects of no gravity, but it's rather hard to study the effects of 1/3rd gravity. Humans might do better than you think. We really don't know.
There is a 100% chance that a species ending asteroid will eventually hit Earth.
No, there's a nonzero chance that when the sun expands past the orbit of Earth, the planet will still be there and will be consumed, and that this will occur before a sufficiently large asteroid hits Earth. And technically, even if the planet has moved to higher orbit and doesn't get consumed, there won't be any species to make extinct at that point, so the asteroid won't be species-ending.
There's also a nonzero chance that we'll manage to blow up the planet somehow, destroying all life before an asteroid hits.
In fact, I doubt the chances are anywhere near 100%. The gas giants, Earth, and Luna have done a fairly decent job of clearing our orbit of most of the larger objects that could hit us.
2. My instinct is that mankind should not be living in large numbers in space. We are big bags of water who are too susceptible to radiation, temperature extremes, pressure drops, lack of oxygen, etc. etc. We are evolved to live on a planetary surface, so Mars it is.
All of those problems would also exist on Mars. It doesn't have a thick enough atmosphere to protect them from ionizing radiation, and a leak would result in an instant loss of pressure. And the temperature on Mars can swing from -100 F to 70 F over the course of a day. If anything, living in space would be easier, because you'd have constant sun exposure on one side and none on the other. Mind you, if we manage to terraform Mars, that's another story, but as-is, it isn't a great match for human life.
Oh please. This has nothing to do with how 'street smart' older people are. What a stupid thing to say. It has to do with fear.
That might be partially true in this specific situation, but the fact remains that the elderly are also much more frequently the victims of many other types of scams that don't involve fear. Pyramid schemes, 419 and sweepstakes scams, fraud involving expensive subscriptions tucked away in the fine print of purchases, people pretending to be family members needing money, etc. all disproportionately affect the elderly.
Now I'll grant you that the elderly are often less likely to report having been scammed out of fear that their kids will get them declared incompetent, and this might make them more popular targets for scammers, but the fact that they are more easily scammed to begin with isn't (necessarily) because of fear.
But there is a problem, and it needs to be recognised and dealt with. There's a serious human cost to removing a whole industry's worth of employment.
I absolutely agree.
We've got to create an upgrade path for those people. Some will want to stay where they are - so we need to encourage some other industry to move in and create employment. Some will want to move to the big city, but once the mining work dries up they won't be able to afford squat.
The people who want to stay where they are are probably mostly screwed. When mines stop being profitable, the towns invariably collapse, and this is likely to have the same result. I don't think that's avoidable. What needs to happen is for someone (government? industry? both?) to provide adequate financial support to help them move somewhere else and retrain as necessary so that they can find work.
The bigger problem, of course, is that this is happening in lots of industries, and eventually there won't be any real call for manual labor. In the short term, house construction is a fairly safe bet, but there aren't enough jobs in that field to accommodate the huge influx of miners, restaurant workers, and factory workers that's coming down the pike. There's always a shortage of nurses, but it isn't for everybody (for lots of reasons) and it requires a considerable amount of intellect, which means not all manual laborers will even have that as an option. Just about any other non-tech, non-creative job that doesn't require a postgraduate degree is likely to be replaced by robots within the next twenty to thirty years.
At some point, we'll have to address that problem—maybe not today, but if we put it off too long, we run the risk of total societal collapse.
It is 2+2 because it was designed for phones. It does not have to 2+2 if defined for a laptop and could be quad-core. Remember Apple ships variants of their Ax processors all the time.
Yeah, but adding cores isn't free. The more cores you add, the more challenging it is to keep their caches in sync. Two cores are relatively easy. Four cores are considerably harder. Six or more cores to match the multicore performance of modern Intel chips are harder still. Obviously it can be done (because it has been done many times), but the point is that cranking up the core count is a non-trivial piece of engineering. The reason it makes sense for Apple to build their own chips for cell phones is because they turn around and build ten million of each model. It makes a lot less sense to spread higher R&D expenses (for a much more complex chip) across a tenth as many devices (or less).
And that's still ignoring the elephant in the room, which is that the per-core speed would need to be half again higher than it currently is, and that's before you consider that initially 100% of the software you run would be emulated. If they could double the per-core speed, it might be practical, but I would be surprised if Apple could reliably clock an A10-derived CPU at 4+ GHz. I mean, you never know, but it doesn't seem very realistic.
Again for a laptop, there would be more room to squeeze in more cores to handle a bigger display.
Although you can generally scale GPUs to massive levels of parallelism much more easily than CPUs, the question becomes whether you're really saving power or expense when you scale up an underpowered GPU by throwing more cores at the problem. And discrete GPUs have been a constant source of heat-induced hardware failures. A scaled-up PowerVR GPU would probably have the same issues unless they specifically designed it to disperse heat evenly. My gut says that a scaled-up design would end up being a ground-up redesign rather than a bunch of existing cores on a single die. I could be wrong, though.
They could, of course rely on discrete GPUs. That's why the GPU performance is less of a problem than the A10's single-core performance. But it is still a significant cost—particularly when you're comparing the board space of a separate GPU versus an Intel chip with an integrated GPU on-die.
On this other hand, at least in this situation, the only words I can think of are "Good. It's about time."
Mining is dangerous work; mines collapse, get filled with dangerous gasses that kill people, and so on. Getting people out of those environments is a great step towards making the world a safer place. I'd imagine their pay will also go down, given that they were getting paid a premium because the job they were doing was dangerous, but that reduction in workers and pay is pretty much unavoidable. The only alternative would be to continue putting people in harm's way unnecessarily, which IMO would be irresponsible once alternatives exist.
I think it's safe to say that they've played with such a configuration. However, unless their R&D is way ahead of what they're shipping, such a device won't actually ship for a long time, if ever.
The biggest problem is that the A10 is still a two-core chip. It has four cores, technically, but two of those are slow cores for reducing power consumption. I don't know if it is possible to use all four cores, but even if it is, it would still be nowhere near as fast as a modern four-core Intel chip. The two fast cores are still only two-thirds to three-quarters the speed of each individual Intel core, and there are half as many, so you would need at least six of those cores just to get the same overall throughput. Worse, CPU-intensive apps like Xcode (where the single-core performance of even current machines is already woefully insufficient) would be utterly unusable on an A10.
The second biggest problem is that the GPU in the A10 is designed to drive a 1920×1080 screen. Even the iPad Pro's GPU is designed to drive only a 2732 x 2048 screen. That's less than half the pixels on an iMac screen, and the iMac's GPU has to routinely drive up to two 3840x2160 UHD screens on top of that. So basically, the GPU performance would probably need to go up by an order of magnitude to replace what's there now, or else they would have to use an external GPU.
I'm not saying it can't be done, and I'm not saying that Apple's engineers couldn't do it, but I don't think computers are important enough to Apple for them to spend the resources required to make it happen. It is far cheaper to just keep cranking out new computers based on Intel's latest designs, and it isn't worth spending that much money for something that produces only 3% of their revenue. I'd expect them to drop the entire Mac product line first.
That's a really lousy reason. It should be possible to solve that problem by using chips with higher density (e.g. by stacking multiple dies in a single epoxy package).
Schiller is technically correct when he says that it is a battery life issue. What he failed to mention is that the only reason that is a problem is because Apple decided that making a pro laptop even more ridiculously thin was more important than making a product that works well for customers. Instead of putting a 100-watt-hour battery in the new MacBook Pro like all the previous models, they chose to reduce the battery capacity by about 25% in the new model to 76 Wh. Because of that decision, the extra ~5W for 32 GB of RAM makes a much bigger difference.
These days, Apple seems to be completely ignoring its customers' needs. Rather than giving us the additional hours of battery life that we have been demanding for years, Apple continued their mindless pursuit of thinness über alles. The result is yet another laptop with less battery life than the previous generation, at least for pro users (whose software is a lot harder on the battery than folks who just use web browsers).
I just don't get it. I really don't. This laptop is what the MacBook should have been. It is the exact opposite of what pro users need, substituting cute gimmicks like a touchscreen top row in place of actual features that pro users need and use regularly, such as battery life, an SD card slot, an HDMI slot, etc. If this is the best Apple can do, then perhaps Michael Dell was right after all.
Not being backwards-compatible for long enough gives you the steaming pile that is running prebuilt binaries on Linux. There's a happy medium. IMO, Apple should have had the foresight to start shipping USB-C chargers and cables with all new iOS devices on the same day....
But this IRS scam is so stupid and transparent that anyone who feel for it has to be so stupid that they really are a menace to society.
If you were talking about people our age, you'd be right. The problem is that these people largely prey on the elderly, many of whom are much less street-smart than folks decades younger. They're from a generation born when these sorts of scams were relatively rare, so if somebody called saying they were from the IRS, they probably were from the IRS—particularly if they have lots of your personal information. It isn't easy to teach college-age folks how to reliably recognize social engineering, much less the elderly. This goes double if they have any sort of mild dementia.
These people prey on the helpless. That makes them utterly inhuman in my book.
For[t] Lauderdale easily has over 1,000 phone rooms going on any given day.
Is it any wonder why they're all in Florida (snowbird central)?
These companies never sell in their own states.
That might have been true in the old days, before they discovered VoIP-based caller ID spoofing. These days, they can sell anywhere from anywhere. All they need is a mailbox somewhere.
Depends. For the OS, apps, and frequently used data, flash is a huge win. For infrequently used, large files like photos, video, etc., spinning storage is fast enough, and is at a better point on the price/performance curve.
Apple is really pushing for you to work over wifi and avoid connecting usb drives.
Then they need to put more than 1 TB of storage in the thing. The days of being able to use a Mac without external storage are in the past, not the future. Five or six years ago, my laptop had 1 TB of storage. Now, even with the hardware they announced today, I'm still stuck with the same capacity, and a lot more data to store in it. I maintain multiple external drives at this point. Apple's hardware hasn't failed to meet my storage needs this badly since the mid-1990s.
One mistake and most buyers will forgive, a second big mistake and they could rapidly lose market share.
Fortunately for Samsung, most buyers have poor memory. I just about gave up on the Blu-Ray format because I had so much trouble with two of their players lasting only a year or so before they stopped reading discs reliably. When I replaced my refrigerator, I seriously considered the Samsung, because (IIRC) its reviews showed ridiculous numbers of people saying that they stopped keeping food cold because of control boards dying (repeatedly)... and a refrigerator is not something you can just toss in the trash can and replace.
This is not to say that they don't also make some products that are good. Obviously they do, or nobody would buy their products. But IMO, it takes a lot more than one or two big mistakes for a company with that much name recognition to rapidly lose market share. Otherwise, it would have happened long ago.:-)
Slashdot Mirror
AFAIK, other than GoPro and cell phones, no cameras use micro-SD. Its a terrible, fragile form factor that is used only in tiny devices that don't have space for a real SD card slot.
That's a showstopper for me, and probably others. With Comodo, I would have to buy a wildcard for hundreds of dollars instead of a few free certs from StartSSL. TLS just went from self-evident to unaffordable and out of reach.
No. All the other free certificates are limited to 90 days. The net effect of this decision is that only big companies and people with too much free time can afford TLS.
No, for the same reason that buying car insurance isn't a waste of money even if there's a chance that you'll never be in a car accident.
Of course they could. My point is that they would end up needing to modify the cores themselves significantly to ramp up the core count, and that those sorts of changes would, I suspect, be too significant to pay off when you're talking about a laptop.
And no, the reason they design their own chips is that they can blow the doors off of what the other companies achieve in terms of power consumption by hand-optimizing the heck out of the core designs. On cell phones, that makes a big difference, and in quantities of tens of millions, the R&D cost per unit is small. On laptops, that makes a much smaller difference (because the batteries are huge by comparison) and the R&D cost per unit is relatively large.
Because nobody will have their software already recompiled for a platform that doesn't exist yet. And the clock speed would need to roughly double if you want emulated code to even approach the same per-core performance that Macs have now. They would need to increase by at least a factor of 1.5 just to get the same per-core performance for native code.
And yet every time I open up my Xcode project at work, Xcode sits there with a single CPU code pegged at 100% for a couple of minutes just to load the project, and several more minutes before it stops SPODing long enough to be usable, and basically the CPU is pegged at 100% for about an hour before indexing finishes. Real-world code doesn't always parallelize easily.
To replace existing machines, the GPU would need to be at least as fast as what they're shipping now....
Or you just plan ahead and mine asteroids to build additional tubes that interconnect. As the population expands, so does the station.
Mars gets almost half as much light as Earth (about 43%, to be more precise). Anything than can grow in partial shade should be able to grow on Mars without the need for artificial light supplementation. Fruits and root vegetables would be more problematic, of course, and would require solar concentrators, but that's still only a 2:1 ratio of ground space versus crop coverage, which isn't *that* insane.
We know the effects of no gravity, but it's rather hard to study the effects of 1/3rd gravity. Humans might do better than you think. We really don't know.
No, there's a nonzero chance that when the sun expands past the orbit of Earth, the planet will still be there and will be consumed, and that this will occur before a sufficiently large asteroid hits Earth. And technically, even if the planet has moved to higher orbit and doesn't get consumed, there won't be any species to make extinct at that point, so the asteroid won't be species-ending.
There's also a nonzero chance that we'll manage to blow up the planet somehow, destroying all life before an asteroid hits.
In fact, I doubt the chances are anywhere near 100%. The gas giants, Earth, and Luna have done a fairly decent job of clearing our orbit of most of the larger objects that could hit us.
All of those problems would also exist on Mars. It doesn't have a thick enough atmosphere to protect them from ionizing radiation, and a leak would result in an instant loss of pressure. And the temperature on Mars can swing from -100 F to 70 F over the course of a day. If anything, living in space would be easier, because you'd have constant sun exposure on one side and none on the other. Mind you, if we manage to terraform Mars, that's another story, but as-is, it isn't a great match for human life.
That might be partially true in this specific situation, but the fact remains that the elderly are also much more frequently the victims of many other types of scams that don't involve fear. Pyramid schemes, 419 and sweepstakes scams, fraud involving expensive subscriptions tucked away in the fine print of purchases, people pretending to be family members needing money, etc. all disproportionately affect the elderly.
Now I'll grant you that the elderly are often less likely to report having been scammed out of fear that their kids will get them declared incompetent, and this might make them more popular targets for scammers, but the fact that they are more easily scammed to begin with isn't (necessarily) because of fear.
I absolutely agree.
The people who want to stay where they are are probably mostly screwed. When mines stop being profitable, the towns invariably collapse, and this is likely to have the same result. I don't think that's avoidable. What needs to happen is for someone (government? industry? both?) to provide adequate financial support to help them move somewhere else and retrain as necessary so that they can find work.
The bigger problem, of course, is that this is happening in lots of industries, and eventually there won't be any real call for manual labor. In the short term, house construction is a fairly safe bet, but there aren't enough jobs in that field to accommodate the huge influx of miners, restaurant workers, and factory workers that's coming down the pike. There's always a shortage of nurses, but it isn't for everybody (for lots of reasons) and it requires a considerable amount of intellect, which means not all manual laborers will even have that as an option. Just about any other non-tech, non-creative job that doesn't require a postgraduate degree is likely to be replaced by robots within the next twenty to thirty years.
At some point, we'll have to address that problem—maybe not today, but if we put it off too long, we run the risk of total societal collapse.
Yeah, but adding cores isn't free. The more cores you add, the more challenging it is to keep their caches in sync. Two cores are relatively easy. Four cores are considerably harder. Six or more cores to match the multicore performance of modern Intel chips are harder still. Obviously it can be done (because it has been done many times), but the point is that cranking up the core count is a non-trivial piece of engineering. The reason it makes sense for Apple to build their own chips for cell phones is because they turn around and build ten million of each model. It makes a lot less sense to spread higher R&D expenses (for a much more complex chip) across a tenth as many devices (or less).
And that's still ignoring the elephant in the room, which is that the per-core speed would need to be half again higher than it currently is, and that's before you consider that initially 100% of the software you run would be emulated. If they could double the per-core speed, it might be practical, but I would be surprised if Apple could reliably clock an A10-derived CPU at 4+ GHz. I mean, you never know, but it doesn't seem very realistic.
Although you can generally scale GPUs to massive levels of parallelism much more easily than CPUs, the question becomes whether you're really saving power or expense when you scale up an underpowered GPU by throwing more cores at the problem. And discrete GPUs have been a constant source of heat-induced hardware failures. A scaled-up PowerVR GPU would probably have the same issues unless they specifically designed it to disperse heat evenly. My gut says that a scaled-up design would end up being a ground-up redesign rather than a bunch of existing cores on a single die. I could be wrong, though.
They could, of course rely on discrete GPUs. That's why the GPU performance is less of a problem than the A10's single-core performance. But it is still a significant cost—particularly when you're comparing the board space of a separate GPU versus an Intel chip with an integrated GPU on-die.
On this other hand, at least in this situation, the only words I can think of are "Good. It's about time."
Mining is dangerous work; mines collapse, get filled with dangerous gasses that kill people, and so on. Getting people out of those environments is a great step towards making the world a safer place. I'd imagine their pay will also go down, given that they were getting paid a premium because the job they were doing was dangerous, but that reduction in workers and pay is pretty much unavoidable. The only alternative would be to continue putting people in harm's way unnecessarily, which IMO would be irresponsible once alternatives exist.
I think it's safe to say that they've played with such a configuration. However, unless their R&D is way ahead of what they're shipping, such a device won't actually ship for a long time, if ever.
The biggest problem is that the A10 is still a two-core chip. It has four cores, technically, but two of those are slow cores for reducing power consumption. I don't know if it is possible to use all four cores, but even if it is, it would still be nowhere near as fast as a modern four-core Intel chip. The two fast cores are still only two-thirds to three-quarters the speed of each individual Intel core, and there are half as many, so you would need at least six of those cores just to get the same overall throughput. Worse, CPU-intensive apps like Xcode (where the single-core performance of even current machines is already woefully insufficient) would be utterly unusable on an A10.
The second biggest problem is that the GPU in the A10 is designed to drive a 1920×1080 screen. Even the iPad Pro's GPU is designed to drive only a 2732 x 2048 screen. That's less than half the pixels on an iMac screen, and the iMac's GPU has to routinely drive up to two 3840x2160 UHD screens on top of that. So basically, the GPU performance would probably need to go up by an order of magnitude to replace what's there now, or else they would have to use an external GPU.
I'm not saying it can't be done, and I'm not saying that Apple's engineers couldn't do it, but I don't think computers are important enough to Apple for them to spend the resources required to make it happen. It is far cheaper to just keep cranking out new computers based on Intel's latest designs, and it isn't worth spending that much money for something that produces only 3% of their revenue. I'd expect them to drop the entire Mac product line first.
That's a really lousy reason. It should be possible to solve that problem by using chips with higher density (e.g. by stacking multiple dies in a single epoxy package).
Schiller is technically correct when he says that it is a battery life issue. What he failed to mention is that the only reason that is a problem is because Apple decided that making a pro laptop even more ridiculously thin was more important than making a product that works well for customers. Instead of putting a 100-watt-hour battery in the new MacBook Pro like all the previous models, they chose to reduce the battery capacity by about 25% in the new model to 76 Wh. Because of that decision, the extra ~5W for 32 GB of RAM makes a much bigger difference.
These days, Apple seems to be completely ignoring its customers' needs. Rather than giving us the additional hours of battery life that we have been demanding for years, Apple continued their mindless pursuit of thinness über alles. The result is yet another laptop with less battery life than the previous generation, at least for pro users (whose software is a lot harder on the battery than folks who just use web browsers).
I just don't get it. I really don't. This laptop is what the MacBook should have been. It is the exact opposite of what pro users need, substituting cute gimmicks like a touchscreen top row in place of actual features that pro users need and use regularly, such as battery life, an SD card slot, an HDMI slot, etc. If this is the best Apple can do, then perhaps Michael Dell was right after all.
Not being backwards-compatible for long enough gives you the steaming pile that is running prebuilt binaries on Linux. There's a happy medium. IMO, Apple should have had the foresight to start shipping USB-C chargers and cables with all new iOS devices on the same day....
If you were talking about people our age, you'd be right. The problem is that these people largely prey on the elderly, many of whom are much less street-smart than folks decades younger. They're from a generation born when these sorts of scams were relatively rare, so if somebody called saying they were from the IRS, they probably were from the IRS—particularly if they have lots of your personal information. It isn't easy to teach college-age folks how to reliably recognize social engineering, much less the elderly. This goes double if they have any sort of mild dementia.
These people prey on the helpless. That makes them utterly inhuman in my book.
Is it any wonder why they're all in Florida (snowbird central)?
That might have been true in the old days, before they discovered VoIP-based caller ID spoofing. These days, they can sell anywhere from anywhere. All they need is a mailbox somewhere.
This latest batch, at least when they got your answering machine, involved bad computer voices (think xtranormal).
Oops, sorry. I missed the mention of Magsafe. :-)
Did the optical output ever appear on the spec sheets?
Depends. For the OS, apps, and frequently used data, flash is a huge win. For infrequently used, large files like photos, video, etc., spinning storage is fast enough, and is at a better point on the price/performance curve.
Good to know. It isn't mentioned on their spec page (or wasn't when I looked right before posting that).
Because it takes courage to give up anonymity.
Even better, like every recent MacBook Pro, the headphone jack has optical outputs. The only thing correct in that post was the amount of RAM.
Then they need to put more than 1 TB of storage in the thing. The days of being able to use a Mac without external storage are in the past, not the future. Five or six years ago, my laptop had 1 TB of storage. Now, even with the hardware they announced today, I'm still stuck with the same capacity, and a lot more data to store in it. I maintain multiple external drives at this point. Apple's hardware hasn't failed to meet my storage needs this badly since the mid-1990s.
Fortunately for Samsung, most buyers have poor memory. I just about gave up on the Blu-Ray format because I had so much trouble with two of their players lasting only a year or so before they stopped reading discs reliably. When I replaced my refrigerator, I seriously considered the Samsung, because (IIRC) its reviews showed ridiculous numbers of people saying that they stopped keeping food cold because of control boards dying (repeatedly)... and a refrigerator is not something you can just toss in the trash can and replace.
This is not to say that they don't also make some products that are good. Obviously they do, or nobody would buy their products. But IMO, it takes a lot more than one or two big mistakes for a company with that much name recognition to rapidly lose market share. Otherwise, it would have happened long ago. :-)
Twitter does more than just run their website. They also do:
Both of those are complex projects involving native code on Android and iOS. Crashlytics also has support for OS X and possibly others.