People love to whine about all the Apple stories. I would defy any of them to submit their own stories about all the other computer companies that are breaking new ground with this type of research. Do you think Dell for example has a team of physics PHDs figuring out these technologies and pushing their vendors to tool up for them? No, THOSE are the guys just packaging off-the-shelf reference designs. Or waiting for the exclusivity on Apple's deal with [insert obscure pacific rim manufacturer here] to expire so they can make a similar looking phone a few years later.
Yes, the guest network can be locked out of the secure private network-- And it works fine behind a router. It just does it's own NAT for it's clients.
Yes I understand the intent of it but I think you missed the crux of the question, which was how does it manage to do _both_ of those things. I.e. how can in be connected to a secure LAN, bridge that through to the secure wireless clients, and still provide a guest network that can NOT access the secure LAN, when the Airport is NOT the sole gateway to the internet.
Sure it would have to NAT the guest network, but then how does it differentiate the secure LAN from the internet? They're both reached on the same interface. You would need a firewall rule. The clean way to do it is to make a VLAN and let the internet gateway handle NAT for both secure AND guest subnets. That is also necessary if you want to serve the same SSIDs from multiple APs for roaming.
I won't even get into DMZ issues but that's basically more of the same problem.
Oh boy, where to begin. First of all there is no such thing as a "half duplex" ethernet cable. One pair is transmit and one pair is receive, unless you're talking about gigabit in which case two pairs are used in each direction.
Secondly, what you describe is NOT PoE. It's a hacky cable with no protection scheme that is likely to cook any other device that you accidentally plug it into.
Thirdly it won't even work in many cases, because CAT5 has a rather high resistance - about 3 per 100 feet. Many devices really need the correct DC voltage supply and if you drop it by a volt or two the input will be too low and not properly regulated.
Proper PoE, that is, 802.3af, uses higher voltages to deal with the resistive losses, and has a detection scheme which prevents damage to non-PoE devices. It also delivers power not by lighting up unused pairs but by putting a bias across the data pairs (which are AC coupled). This means you don't use extra pairs and it can still work with gig-E.
Proper PoE switches/injectors and "splitters" are now cheap enough that there's not much reason to go fiddling with special cables and risking damage to your equipment.
I don't disagree with you. I have not yet tried the airport extreme, as I managed to resolve my major linksys problem by adding a fan. But I still have problems with Mac clients occasionally failing to reconnect when roaming and/or waking from sleep and I would hope that's not a problem if the APs are from the same vendor.
Do you know if the guest mode would work in the case where another router (Cisco ASA) is doing NAT to the internet? It would be important that the guest network only be able to access public IPs and not the secured LAN, but I'm not clear in that case how it would know the difference if it's assuming a typical home network with a directly connected WAN. I would be surprised if it lets me assign the guest network to a VLAN that I define, but that would be ideal. In any case it's a feature I can live without - as I was saying, a reliable AP is the only must-have.
Well in the case of my 350n I cut an opening in the top of the case to reveal the (totally encapsulated in plastic) shield / heat sink. Then mounted an 80mm fan on a little bracket 2 cm away. That solved the problem. But mine was in a well ventilated room temperature space - in an attic you're probably going to have problems in the summer no matter what. Maybe put your whole installation (I'm assuming you have more equipment) in an enclosure and use a bathroom exhaust fan to push cooler outdoor air through it. Or run in/out ducts into your living space (eg vent through the ceiling in a closet) if that's cooler.
The perfect 802.11n router for me is the one that just acts as a reliable AP and doesn't overheat, crash, drop connections, or have special compatibility problems. How about making it WORK before you add more "value"?
DDWRT helps but the hardware on the market is just garbage. And it's NOT because it's made of commodity components, but because it's poorly engineered. Best example of this is the horrific power/thermal management on newer Linksys products. Ethernet _switch_ traffic alone is enough to make the whole system overheat and crash no matter what firmware you're running. A competent engineer could have made it work right for the same BOM. I used to make wireless devices and our biggest category of support problems was crappy wireless routers either spontaneously rebooting, or needing to be rebooted. I just can't believe we are still at the same state of reliability as the 802.11b days - actually it seems worse now.
PS I don't mean to pick on Linksys, it's just that they're the ones I'm most familiar with. Overall the fails seemed to be in proportion to market share although every one had its particular problems.
You're probably making too high of an assumption about the incremental cost to add this to consumer products. I would be surprised if it's even a single square mm of die area. All depends how they price the IP.
You might think a background image is nothing worth getting in a huge flap over, but let me explain why this experiment was so stupid.
First, Google seems to have forgotten the early days of the search engine wars in which Yahoo, Excite, et al vied for the most user-hostile, craptacular portal landing pages. I believe it was primarily their choice of a minimal utilitarian design that made people flock to Google, and the quality of the search results, good as they were, was a distant secondary factor among typical users.
Secondly, the actual execution of this feature was terrible. Not only were the images bright, garish, and distracting, but there was NO option to turn it off. Sure if you spent a few minutes digging you could find the "editor's choice" images, and if you scrolled all the way down to the bottom you could find white. But then if you picked that, you would get white text on a white background. Brilliant.
Google has said in the past that they use an empirical, incremental approach to UI design where user actions are studied and these guide decisions down to the level of how many pixels to make a line or what font size to use. Some have rightly pointed out that this will cause you to get stuck on local maxima and you need to have a methodology that allows for some creative design. But forcing such a butt ugly intrusion on all users for the purposes of a trial is ridiculous. If they really wanted to do a trial they could have simply served this to, say, 1 in 10,000 users (based on IP+useragent hash, for example) and got the exact same information.
No, this could only have been the brainchild of a marketroid who thought it would be necessary to "make a splash" and get some "buzz" going. Well congratulations, you got your feedback and the answer is a resounding "fuck off". Google has officially run out of ideas if this is the best they can come up with.
And if there is money to be had, those businesses will scale up to meet your demands, eventually.
Or not. Domestic suppliers will interpret rising volume as an indicator that you will soon be drawn offshore for lower pricing. Unless they have other customers who will use the facilities, they will not scale their production capital just to watch you jump ship. When it comes to electronics they really only cater to the small volume / high margin production that is still feasible in the US.
In the longer term that might change with the failure of the USD, but the network effect of the Asian supply chain that you mention is going to make it a difficult transition.
Perhaps I might have asked _why_ it was modded as such.
I have mod points and I get to assign them anyway that I want.
You can certainly TRY, but meta-moderation will sort that out and if you keep it up the system will just stop giving you mod points. "off topic" is not a catch-all for "I don't like this".
They expect everyone to sign up for their "publishing environment" to add these embedded codes to their books? How long is THAT going to take before there's any critical mass of dead trees in people's hands to be able to use this?
Google and Amazon already have thousands of books scanned. All they would need is a photo of any _existing_ book page, do a ballpark OCR on it and fuzzy match the database.
So even if this is a useful idea, which I'm not seeing, from a practical standpoint they are never going to get off the ground with this approach. And if some compelling application were ever found, they would be crushed by the guys with the books.
So a note to the inventors - ditch this ridiculous scheme and if you think there's an opportunity here, figure out how to make it work with the books we already have.
Fair enough, although I hardly think that having extracted a couple %/yr of dividends up to that time would bring much consolation to the shareholders who lost their principal entirely.
For the other ones you mention I suppose I could have qualified my statement to exclude cases gross criminal fraud.
You either have no idea how equities actually work or you are being deliberately obtuse. Do you think huge companies just instantly vanish in some random event that nobody can predict? No. All that cash in the bank? It belongs to the shareholders. Likewise all the other assets of the company. You can take profits selling your AAPL any time you like - right now it's close to its all-time-high.
The reason fast growing companies don't pay dividends is because shareholders don't want them. The reasons they don't want them are:
1) Dividends are taxed as ordinary income, which by the time you figure typical state taxes is about double the rate for long term cap gains.
2) If the company has a use for the money (eg mergers, or even just stockpiling for future downturns) then it's better for the shareholders to keep the cash in the corporation than to be forced to raise money through borrowing, or dilution at times when the stock is weaker.
3) Shareholders want to choose when they take their profits/income.
The shareholders literally, legally, OWN the company. It is more than simply a game of betting which way the share price will move although many people look at it that way.
When you look at an established company with good profitability but shaky prospects for future growth - case in point, MSFT, that is when shareholders will demand a dividend. If the stock isn't going up and the company can't find efective ways to reinvest its profits, then the shareholders will want a periodic return. AAPL shareholders are expecting more from the company right now and you'd have been a fool to bet against them at any time in the last 12 years or so. That's not saying it goes up forever but AAPL has so much cash and such a broad product portfolio that it's not the kind of thing that can go out of business without a seriously protracted unwinding.
And your notion that it's only a matter of having a sufficiently long time frame is patently false. Never mind that this only can happen if the entirety of the company's assets are liquidated to satisfy its debts... what if they are acquired in whole? I suppose you would argue that the acquiring entity eventually will fail, etc ad infinitum. Well that's entropy for ya - heat death of the universe. So what? Should nobody invest then? Should successful growing companies all pay dividends instead of using that money to expand operations?
Why don't you invest in such companies while the rest of us hold our AAPL. Sounds like you've got an angle!
During the dot-com boom, as I recall there was no shortage of customers for data centers, and every one I visited was filling up new space as fast as they could equip it. Mostly with expensive servers that were underutilized. The problem was those customers were ultimately not viable. They weren't building "on spec".
Still I agree that this rising demand on the tail of the recession is a good sign, for the valley in particular.
All that's going to happen is an increase in your petrol prices (orchestrated together by all oil companies)
I for one would rather pay for my gas per volume at the pump, rather than in income taxes, budget deficits, and inflation. Just add up all the wars, toxic waste cleanups, corporate welfare, the whole fucking nine yards and bill it right there at the pump instead of on my tax return. While you're at it, throw in road maintenance.
You didn't answer his question at all, which I thought was a good one. He said nothing of earning a living but rather to "make interesting discoveries and potentially chart new territory in the home". Well, he did say "make an investment" but I read that as "in myself / my hobby".
I don't know the answer. The areas of science that I could imagine practicing at home are well trodden. That's not going to stop me from making electromechanical things for fun, but I don't expect to change the the world with it.
What the hell is the "white phase"? Unless I am missing some newfangled data-center lingo, you are talking about the neutral, which is not a "phase" at all, and could never produce such a fault current when "shorted" to ground since it is already tied to ground at the panel. Am I missing something?
Umm, starting your posts with "umm" makes you look awfully clueless when you're the one who is STILL not getting it. Yes you can do -almost- whatever you want with BSD code, including distributing proprietary software without source. However, you CAN NOT claim it as your own. It must retain a BSD license notice identifying the original authors. I would suggest you actually read the license before commenting further, as this is the central point of it.
Apple can take a BSD programmer's code, and claim it as their own.
Actually, that is the one thing you CAN'T do with BSD code. The attribution requirements are practically the only difference between BSD and public domain.
we would still be using floppy disks and parallel ports. Even if you don't like their products, or don't recognize this as progress, I see no reason to be so snide about it.
Gettng Started in Electronics. It takes you through everything from basic soldering to building logic circuits, oscillators, amplifiers. His "mini notebooks" are great too.
Once you have the basics down you will probably want to get into microcontrollers. There are a lot of ways to go here depending on how much time you want to spend wiring things up yourself, and your comfort level with software. You might start with the very popular PIC. Although the architecture is a bit long in the tooth and is a poor target for C, there loads of example projects for it so it's easy to learn. There are also many high-level building blocks (Basic stamp etc) that can get you up and running quickly. If you have sophisticated software needs, you'll want a more modern micro with better tools - check out Atmel or TI.
Eventually you will need a more formal treatment if you want to design your own circuits. I consider The Art of Electronics to be the bible here - it is thorough but also very practical and you will find it has specific solutions for many everyday engineering problems. It has been a great investment, and one of the better worn books on my shelf. Have fun!
Slashdot Mirror
People love to whine about all the Apple stories. I would defy any of them to submit their own stories about all the other computer companies that are breaking new ground with this type of research. Do you think Dell for example has a team of physics PHDs figuring out these technologies and pushing their vendors to tool up for them? No, THOSE are the guys just packaging off-the-shelf reference designs. Or waiting for the exclusivity on Apple's deal with [insert obscure pacific rim manufacturer here] to expire so they can make a similar looking phone a few years later.
Yes I understand the intent of it but I think you missed the crux of the question, which was how does it manage to do _both_ of those things. I.e. how can in be connected to a secure LAN, bridge that through to the secure wireless clients, and still provide a guest network that can NOT access the secure LAN, when the Airport is NOT the sole gateway to the internet.
Sure it would have to NAT the guest network, but then how does it differentiate the secure LAN from the internet? They're both reached on the same interface. You would need a firewall rule. The clean way to do it is to make a VLAN and let the internet gateway handle NAT for both secure AND guest subnets. That is also necessary if you want to serve the same SSIDs from multiple APs for roaming.
I won't even get into DMZ issues but that's basically more of the same problem.
Secondly, what you describe is NOT PoE. It's a hacky cable with no protection scheme that is likely to cook any other device that you accidentally plug it into.
Thirdly it won't even work in many cases, because CAT5 has a rather high resistance - about 3 per 100 feet. Many devices really need the correct DC voltage supply and if you drop it by a volt or two the input will be too low and not properly regulated.
Proper PoE, that is, 802.3af, uses higher voltages to deal with the resistive losses, and has a detection scheme which prevents damage to non-PoE devices. It also delivers power not by lighting up unused pairs but by putting a bias across the data pairs (which are AC coupled). This means you don't use extra pairs and it can still work with gig-E.
Proper PoE switches/injectors and "splitters" are now cheap enough that there's not much reason to go fiddling with special cables and risking damage to your equipment.
I don't disagree with you. I have not yet tried the airport extreme, as I managed to resolve my major linksys problem by adding a fan. But I still have problems with Mac clients occasionally failing to reconnect when roaming and/or waking from sleep and I would hope that's not a problem if the APs are from the same vendor. Do you know if the guest mode would work in the case where another router (Cisco ASA) is doing NAT to the internet? It would be important that the guest network only be able to access public IPs and not the secured LAN, but I'm not clear in that case how it would know the difference if it's assuming a typical home network with a directly connected WAN. I would be surprised if it lets me assign the guest network to a VLAN that I define, but that would be ideal. In any case it's a feature I can live without - as I was saying, a reliable AP is the only must-have.
Well in the case of my 350n I cut an opening in the top of the case to reveal the (totally encapsulated in plastic) shield / heat sink. Then mounted an 80mm fan on a little bracket 2 cm away. That solved the problem. But mine was in a well ventilated room temperature space - in an attic you're probably going to have problems in the summer no matter what. Maybe put your whole installation (I'm assuming you have more equipment) in an enclosure and use a bathroom exhaust fan to push cooler outdoor air through it. Or run in/out ducts into your living space (eg vent through the ceiling in a closet) if that's cooler.
DDWRT helps but the hardware on the market is just garbage. And it's NOT because it's made of commodity components, but because it's poorly engineered. Best example of this is the horrific power/thermal management on newer Linksys products. Ethernet _switch_ traffic alone is enough to make the whole system overheat and crash no matter what firmware you're running. A competent engineer could have made it work right for the same BOM. I used to make wireless devices and our biggest category of support problems was crappy wireless routers either spontaneously rebooting, or needing to be rebooted. I just can't believe we are still at the same state of reliability as the 802.11b days - actually it seems worse now.
PS I don't mean to pick on Linksys, it's just that they're the ones I'm most familiar with. Overall the fails seemed to be in proportion to market share although every one had its particular problems.
You're probably making too high of an assumption about the incremental cost to add this to consumer products. I would be surprised if it's even a single square mm of die area. All depends how they price the IP.
First, Google seems to have forgotten the early days of the search engine wars in which Yahoo, Excite, et al vied for the most user-hostile, craptacular portal landing pages. I believe it was primarily their choice of a minimal utilitarian design that made people flock to Google, and the quality of the search results, good as they were, was a distant secondary factor among typical users.
Secondly, the actual execution of this feature was terrible. Not only were the images bright, garish, and distracting, but there was NO option to turn it off. Sure if you spent a few minutes digging you could find the "editor's choice" images, and if you scrolled all the way down to the bottom you could find white. But then if you picked that, you would get white text on a white background. Brilliant.
Google has said in the past that they use an empirical, incremental approach to UI design where user actions are studied and these guide decisions down to the level of how many pixels to make a line or what font size to use. Some have rightly pointed out that this will cause you to get stuck on local maxima and you need to have a methodology that allows for some creative design. But forcing such a butt ugly intrusion on all users for the purposes of a trial is ridiculous. If they really wanted to do a trial they could have simply served this to, say, 1 in 10,000 users (based on IP+useragent hash, for example) and got the exact same information.
No, this could only have been the brainchild of a marketroid who thought it would be necessary to "make a splash" and get some "buzz" going. Well congratulations, you got your feedback and the answer is a resounding "fuck off". Google has officially run out of ideas if this is the best they can come up with.
Or not. Domestic suppliers will interpret rising volume as an indicator that you will soon be drawn offshore for lower pricing. Unless they have other customers who will use the facilities, they will not scale their production capital just to watch you jump ship. When it comes to electronics they really only cater to the small volume / high margin production that is still feasible in the US.
In the longer term that might change with the failure of the USD, but the network effect of the Asian supply chain that you mention is going to make it a difficult transition.
You can certainly TRY, but meta-moderation will sort that out and if you keep it up the system will just stop giving you mod points. "off topic" is not a catch-all for "I don't like this".
Who modded this offtopic? I hope that was an error as this is most certainly to the point.
Google and Amazon already have thousands of books scanned. All they would need is a photo of any _existing_ book page, do a ballpark OCR on it and fuzzy match the database.
So even if this is a useful idea, which I'm not seeing, from a practical standpoint they are never going to get off the ground with this approach. And if some compelling application were ever found, they would be crushed by the guys with the books.
So a note to the inventors - ditch this ridiculous scheme and if you think there's an opportunity here, figure out how to make it work with the books we already have.
Fair enough, although I hardly think that having extracted a couple %/yr of dividends up to that time would bring much consolation to the shareholders who lost their principal entirely. For the other ones you mention I suppose I could have qualified my statement to exclude cases gross criminal fraud.
Not sure if serious. If you are, citation please and I will rebut.
The reason fast growing companies don't pay dividends is because shareholders don't want them. The reasons they don't want them are:
1) Dividends are taxed as ordinary income, which by the time you figure typical state taxes is about double the rate for long term cap gains.
2) If the company has a use for the money (eg mergers, or even just stockpiling for future downturns) then it's better for the shareholders to keep the cash in the corporation than to be forced to raise money through borrowing, or dilution at times when the stock is weaker.
3) Shareholders want to choose when they take their profits/income.
The shareholders literally, legally, OWN the company. It is more than simply a game of betting which way the share price will move although many people look at it that way.
When you look at an established company with good profitability but shaky prospects for future growth - case in point, MSFT, that is when shareholders will demand a dividend. If the stock isn't going up and the company can't find efective ways to reinvest its profits, then the shareholders will want a periodic return. AAPL shareholders are expecting more from the company right now and you'd have been a fool to bet against them at any time in the last 12 years or so. That's not saying it goes up forever but AAPL has so much cash and such a broad product portfolio that it's not the kind of thing that can go out of business without a seriously protracted unwinding.
And your notion that it's only a matter of having a sufficiently long time frame is patently false. Never mind that this only can happen if the entirety of the company's assets are liquidated to satisfy its debts... what if they are acquired in whole? I suppose you would argue that the acquiring entity eventually will fail, etc ad infinitum. Well that's entropy for ya - heat death of the universe. So what? Should nobody invest then? Should successful growing companies all pay dividends instead of using that money to expand operations?
Why don't you invest in such companies while the rest of us hold our AAPL. Sounds like you've got an angle!
Still I agree that this rising demand on the tail of the recession is a good sign, for the valley in particular.
I for one would rather pay for my gas per volume at the pump, rather than in income taxes, budget deficits, and inflation. Just add up all the wars, toxic waste cleanups, corporate welfare, the whole fucking nine yards and bill it right there at the pump instead of on my tax return. While you're at it, throw in road maintenance.
Then let's see how we feel about conserving oil.
I don't know the answer. The areas of science that I could imagine practicing at home are well trodden. That's not going to stop me from making electromechanical things for fun, but I don't expect to change the the world with it.
The hots are black, red, and blue (in that order of prevalence) in the US.
What the hell is the "white phase"? Unless I am missing some newfangled data-center lingo, you are talking about the neutral, which is not a "phase" at all, and could never produce such a fault current when "shorted" to ground since it is already tied to ground at the panel. Am I missing something?
Umm, starting your posts with "umm" makes you look awfully clueless when you're the one who is STILL not getting it. Yes you can do -almost- whatever you want with BSD code, including distributing proprietary software without source. However, you CAN NOT claim it as your own. It must retain a BSD license notice identifying the original authors. I would suggest you actually read the license before commenting further, as this is the central point of it.
Actually, that is the one thing you CAN'T do with BSD code. The attribution requirements are practically the only difference between BSD and public domain.
we would still be using floppy disks and parallel ports. Even if you don't like their products, or don't recognize this as progress, I see no reason to be so snide about it.
Once you have the basics down you will probably want to get into microcontrollers. There are a lot of ways to go here depending on how much time you want to spend wiring things up yourself, and your comfort level with software. You might start with the very popular PIC. Although the architecture is a bit long in the tooth and is a poor target for C, there loads of example projects for it so it's easy to learn. There are also many high-level building blocks (Basic stamp etc) that can get you up and running quickly. If you have sophisticated software needs, you'll want a more modern micro with better tools - check out Atmel or TI.
Eventually you will need a more formal treatment if you want to design your own circuits. I consider The Art of Electronics to be the bible here - it is thorough but also very practical and you will find it has specific solutions for many everyday engineering problems. It has been a great investment, and one of the better worn books on my shelf. Have fun!
What is the guy behind the building doing at time 4:08?