This is probably a highly unpopular point of view, but I sympathise with Microsoft on this one.
There is no clear definition of what an operating system is. DOS, for instance, was little more than a program loader (similar to what used to be called a monitor) plus a simple shell. Other OSs provide task management, privilege management, and so on. All three main desktop OSs now current go further than this to provide user interaction elements. These are most clear in the case of the windowing system, providing menus, dialogs and so on. A part of what they do is to provide rendering: for instance MacOS has native support for a form of Postscript, and Windows does the same for WMF. Both of these are useful, but I'd say that it is more important for an OS to provide native support for the dominant rendering language, HTML. Windows provides this: there is a controls which can be embedded in to a third party application to allow it to render HTML in a consistent way (albeit there are some non-standard behaviours). Similarly, it does make sense for an OS to provide some higher level network communication protocols - FTP, HTTP and so on. In other words, I think it does make sense for any modern OS to provide the components of a web browser, and MS were wise in componentising IE before Netscape.
From this position of supplying the components, it is relatively easy to bolt together a complete web browser (I have heard that ie.exe is only 80kb, but I don't have a Windows PC to hand to check this). Put yourself in the position of a technical manager in MS - you've done all the hard work to provide the components. If you do the simple task of integration, your users get a web browser. If you don't, you're providing an OS which out of the box cannot interface to the dominant means of Internet communication, and can't be bootstrapped up by downloading other browsers.
Ok, I admit that I'm only presenting one side of the case here, but my point is that MS do have some justification for their design decisions. I sometimes have to deal with MS development plans, and my impression is that there's nothing evil about their mid-level management - it's just that they are a monoculture and don't consider the effects of their actions on the rest of the software ecosystem.
Objective C uses reference counting, not garbage collection - they're not the same thing. RC requires some manual work to maintain the counts, introducing some possibility of error, and it can be broken if you use structures such as circular lists. Several studies have also shown it to be slower than GC (I'm not entirely convinced myself).
Probably obvious to you, but for those who don't know - SIP is the "session initiation protocol". It's just the call setup mechanism, but it specifies what form the call will take. The most common type of call is over RTP (real time protocol), but this isn't mandatory. RTP is based on UDP, which means that you can't trivially use TLS (which requires the reliable transport of TCP to keep in sequence). Two means of encryption have been proposed to secure the call voice stream (as opposed to the setup data, which is the main subject of the articial. These are SRTP and DTLS, the latter being a generic UDP encryption mechanism which is "heavier" than SRTP. Last time I looked at these, about 18 months ago, there were no problem-free implementations generally available, but this may have changed. Of course if you were sufficiently determined, you could implement your own stack to implement DTLS or SRTP.
The main problem in design is not so much the overhead of the encryption, which is usually bearable, but the secure setup of the encryption key to be used for SRTP or DTLS. NAT in particular is a killer. TLS for web sites relies on DNS and public IP addresses to work with the certificates. You can't normally use this where one or both of the call parties are on private non-fixed IP addresses (there is a partial exception if they are on the same private network [no NAT between them], they have each been issued with certificates, and something is tying DHCP and DNS together).
The situation is also complicated by needing two-way authentication (TLS usually only authenticates the server, although client authentication can also be done). In the SIP case, one part of the problem is that the SIP session has authenticated both users, but we have to export this information so that the RTP session knows that they have been authenticated.
There are a couple of proposed means of key exchange, but when I last looked at this one was only an Internet Draft, and the other had gone to RFC a couple of weeks earlier - neither very mature, so very difficult to do a security analysis on them.
Interestingly, this would probably be illegal in the UK under Part II of the "big brother" RIPA (Regulation of Investigatory Powers Act). Usually we hear about it as an egregious violation of privacy, but on the one occasion I've had to deal with it, I was able to tell a US business partner that there was no way I was going to add in a monitor that they wanted, and give them chapter and verse.
First? Not by more than ten years. Psions used to start up faster than you could open the lid. I still wonder what Macs and Windows machines are *doing* with all that time.
[I work for one of the major mobile telcos in a group that deals with mobile data clients]
Yes, it's perfectly possible to get the device to use EDGE as well as 3G and many devices do that. However very few networks upgraded their GSM equipment to use EDGE as they went straight to 3G so in practice I've only ever seen this when I visited the USA and roamed on to the Cingular network.
While I don't have the figures, as far as I know GPRS/EGPRS (i.e. EDGE GPRS) has much higher latency than UMTS, HSDPA or HSUPA. The reason is that it uses time slicing to get multiple user on to a single frequency, whereas WCDMA (the radio technology that UMTS et al. are based on) allows simultaneous transmission. The max throughput is hugely different: for instance I have a pretty common handset that theoretically does 3.6Mb/s, and I get something close to that in practice. 7.2Mb/s is coming soon, if not already out (I'm not up to date with the network plans). The only area that the older technologies are better on is power consumption: there's a big difference on this, but if you need to eke out your battery, most handsets will allow you to select GSM/GPRS only.
One they really should mention because it usually catches new Mac users out: how to de-install an application. I spent some time looking for the Add/Remove Programs equivalent before I found that you just drop the application in the Trash.
(For those who don't know, Currys and PC World are branches of the same company)
I'd support that: I buy a lot of stuff at PC World precisely because they've been utterly reliable on returns - they may be a bit more expensive than the web for some stuff, but it's been worth it.
No, no-one has reported cracking it. Bear in mind that Ken is capable of hiding stuff below the source code of the OS, Ken could have set it up so that when a program outputs this particular string, Unix takes some predetermined action such as calling in the black helicopters.
On a more serious level, but for the same reason, there is no reason to think that this entry in the password file corresponds to a valid Unix password, since if that system was based on his code, he login will bypass normal authentication completely.
If you don't understand what I'm talking about - have a look at the paper. It's a classic, and well worth a read. Wikipedia has a summary .
There seem to be a lot of comments saying that this device is useless to artists because of lack of resolution and pressure sensitivity. True, but it misses the point. I used to use a Psion 7, a lovely little A5 size clamshell machine running a proprietary OS. It was ideal for taking notes in meetings as I could go straight from touch-typing to sketching a diagram straight into the word processor. These days I use MS OneNote, which is slightly more clumsy in that respect (much better in other ways). If I have a tablet I can sketch into my notes, but usually it's too much hassle to take the tablet and find room on a table in a crowded meeting room. A device that clipped on to the screen would help to a certain extent. The Psion would still be better in having the hw built in, internal stowage for the stylus, and a fancy hinge that brought the bottom of the screen forward to avoid the machine tipping backwards when the touch screen was used.
Re:US fuel efficiency figures seem incredibly poor
on
Japanese Mileage Maniacs
·
· Score: 2, Informative
The US gallon is only 80% of the imperial gallon - it's not clear which units they are using. FWIW, I've got 47mpg (imperial) over 25000 miles in my Jaguar X-Type diesel without doing more than roughly keep to the speed limits, so I'm not overly impressed whichever units are in use.
The ISP I'm with seems to have a fair policy. They only meter 08:00-18:00 Monday to Friday. There are tariffs offering 1, 3 5 and 8Gb of traffic during these hours, unlimited outside. If you go over, their policy states (in part):
What if I use too much?
Don't panic! As always, we have been careful to design tariffs that have limits a typical user will never reach. However, if you do reach your limit we will email you. You can change the level at which you get this email. We do not automatically cut you off or slow down your service. For fixed usage tariffs you can carry over up to 10GB from one month to the next without any penalty, so a really heavy month can be caught up over several future months. If you have used more than this, then you will be contacted so you can pay for top-up and/or select a more appropriate tariff. If you are being unreasonable, we can suspend your service, but that is always a last resort. Customers (typically businesses) can also opt to be billed for usage automatically in arrears if they wish. For variable usage tariffs we charge the extra usage in 0.01GB steps, so you only pay for the extra amount of usage you actually used.
Seems simple, fair and workable to me. There's other reasons I like them - my block of fixed public IP addresses at no extra charge, and the continuous ping-testing of my line so that they SMS me if it goes down.
TGV-type trains work best over fairly long distance without stops - for instance Paris to Marseille is 500 miles in 3 hours. I frequently have to travel a similar distance by plane, and although the time in the air might be about half that, door to door the train would be faster because of the overheads at the airports.
People from Asian or European countries just don't understand how much space lies between American cities.
Yes, we do. We're just not that impressed by it. Oh, and have you seen the size of Russia, which last time I looked was mainly an Asian country?
The 3 was a nice machine to hold, but I'd go for an updated Series 7 / Netbook: A5 format; big enough to touch type notes in real time; that fancy cantilevered hinge bringing the bottom of the screen forward so that it didn't fall over backwards when you started drawing a diagram in the middle of a document (why doesn't any modern laptop have this?); wrap around leather so that it felt good to carry folded in the hand. Also unlike the 3 it didn't jettison its batteries when you dropped it: it was rated for a 1m fall onto concrete.
Update that with a modern display, WiFi, Bluetooth and HSUPA (UMTS is so two years ago) and you've got my dream machine. As to software: I've become very fond of M$ OneNote - I'd like something like that on it.
Sounds like a misunderstanding. IPv6 addresses are hierarchical. A/32 would be allocated to an ISP, and you should get a/48 from them (yes, I've done this). If your upstream ISP doesn't distribute IPv6 addresses, they aren't going to be able to route IPv6 either, so you need to find a tunnel broker. Any tunnel broker will give you a range, either a/48 or a/64, which you can use with a fixed tunnel. Alternatively you can set up a 6to4 tunnel using the anycast addresses 192.88.99.1 and 2002:c058:6301:: as the far end, which will give you a unique/48 based on the IPv4 address of your router - however if you're serious about building a commercial network on IPv6, you should probably go for the fixed link.
I've been through Kiruna a few times to go walking. It's a big, sparsely populated mining and forestry town in the Saami (Lapp) area of the north of Sweden. The air connection has to be subsidised by the government, and it's a long flight from Stockholm Arlanda in a very small plane. The air crew come round to ask who would like a taxi called for them at the airport. When you arrive, there's a single small luggage carousel and a large stuffed bear in Arrivals.
A friend of mine, son of a judge, remarked that the legal term "the third head of malice" would be an excellent name for a heavy metal band.
Since common law rests on precedent, old judgements are often quoted verbatim. One I particularly liked contained the phrase "every cyclist is entitled to his wobble".
I think you misunderstand what these devices are doing. They are proxies for the edge of a corporate network, and they assume that the IT manager explicitly configures the proxy settings for his users' web browsers to use the proxy (and blocks any direct Internet access using the firewall). There is an SSL/TLS connection from the browser to the proxy, and from the proxy to the target web site. They do act as a man in the middle, but this isn't in any way remarkable as it can only happen with an explicit request from the web browser.
Probably not. They burn something closer to axle grease than to oil, but it's pre-heated before injection. I've seen a similar five-cylinder engine, which used two turbochargers of 3 foot diameter, which would tend to raise the air temperature despite the use of intercoolers. In addition to those points, with cylinders that size the wall temperature isn't going to affect temperature through most of the volume of the combustion chamber simply because most of the gas isn't in contact with the wall.
On similar lines but with less detail, I've been wondering whether it would be a good idea to allocate different IP addresses to different processes. IPv6 has enough address space to make this feasible. Currently personal firewalls (Zone Alarm, Little Snitch et al.) do a reasonable job of stopping spyware, but they have to run on the same machine as the processes they are guarding. If processes had different IP addresses on a single host, it would allow external conventional firewalls to take over a large part of this role, and would also address per-application packet shaping. You'd need something like DHCP to hand out addresses to processes, possibly based on a combination of the program (SHA hash?) and the user executing it.
It would potentially allow a network/fw admin to set policies such as "Internet Explorer may only be used to browse intranet sites"; "These versions of Firefox may be used to browse externally".
In general, you can't even send data (other than SMS) simultaneously with voice. The standards allow it, but it isn't usually implemented. Not sure how true this is for 3G phones, though.
Slashdot Mirror
The Harrier. springs to mind.
There is no clear definition of what an operating system is. DOS, for instance, was little more than a program loader (similar to what used to be called a monitor) plus a simple shell. Other OSs provide task management, privilege management, and so on. All three main desktop OSs now current go further than this to provide user interaction elements. These are most clear in the case of the windowing system, providing menus, dialogs and so on. A part of what they do is to provide rendering: for instance MacOS has native support for a form of Postscript, and Windows does the same for WMF. Both of these are useful, but I'd say that it is more important for an OS to provide native support for the dominant rendering language, HTML. Windows provides this: there is a controls which can be embedded in to a third party application to allow it to render HTML in a consistent way (albeit there are some non-standard behaviours). Similarly, it does make sense for an OS to provide some higher level network communication protocols - FTP, HTTP and so on. In other words, I think it does make sense for any modern OS to provide the components of a web browser, and MS were wise in componentising IE before Netscape.
From this position of supplying the components, it is relatively easy to bolt together a complete web browser (I have heard that ie.exe is only 80kb, but I don't have a Windows PC to hand to check this). Put yourself in the position of a technical manager in MS - you've done all the hard work to provide the components. If you do the simple task of integration, your users get a web browser. If you don't, you're providing an OS which out of the box cannot interface to the dominant means of Internet communication, and can't be bootstrapped up by downloading other browsers.
Ok, I admit that I'm only presenting one side of the case here, but my point is that MS do have some justification for their design decisions. I sometimes have to deal with MS development plans, and my impression is that there's nothing evil about their mid-level management - it's just that they are a monoculture and don't consider the effects of their actions on the rest of the software ecosystem.
Objective C uses reference counting, not garbage collection - they're not the same thing. RC requires some manual work to maintain the counts, introducing some possibility of error, and it can be broken if you use structures such as circular lists. Several studies have also shown it to be slower than GC (I'm not entirely convinced myself).
The main problem in design is not so much the overhead of the encryption, which is usually bearable, but the secure setup of the encryption key to be used for SRTP or DTLS. NAT in particular is a killer. TLS for web sites relies on DNS and public IP addresses to work with the certificates. You can't normally use this where one or both of the call parties are on private non-fixed IP addresses (there is a partial exception if they are on the same private network [no NAT between them], they have each been issued with certificates, and something is tying DHCP and DNS together).
The situation is also complicated by needing two-way authentication (TLS usually only authenticates the server, although client authentication can also be done). In the SIP case, one part of the problem is that the SIP session has authenticated both users, but we have to export this information so that the RTP session knows that they have been authenticated.
There are a couple of proposed means of key exchange, but when I last looked at this one was only an Internet Draft, and the other had gone to RFC a couple of weeks earlier - neither very mature, so very difficult to do a security analysis on them.
Interestingly, this would probably be illegal in the UK under Part II of the "big brother" RIPA (Regulation of Investigatory Powers Act). Usually we hear about it as an egregious violation of privacy, but on the one occasion I've had to deal with it, I was able to tell a US business partner that there was no way I was going to add in a monitor that they wanted, and give them chapter and verse.
First? Not by more than ten years. Psions used to start up faster than you could open the lid. I still wonder what Macs and Windows machines are *doing* with all that time.
Yes, it's perfectly possible to get the device to use EDGE as well as 3G and many devices do that. However very few networks upgraded their GSM equipment to use EDGE as they went straight to 3G so in practice I've only ever seen this when I visited the USA and roamed on to the Cingular network.
While I don't have the figures, as far as I know GPRS/EGPRS (i.e. EDGE GPRS) has much higher latency than UMTS, HSDPA or HSUPA. The reason is that it uses time slicing to get multiple user on to a single frequency, whereas WCDMA (the radio technology that UMTS et al. are based on) allows simultaneous transmission. The max throughput is hugely different: for instance I have a pretty common handset that theoretically does 3.6Mb/s, and I get something close to that in practice. 7.2Mb/s is coming soon, if not already out (I'm not up to date with the network plans). The only area that the older technologies are better on is power consumption: there's a big difference on this, but if you need to eke out your battery, most handsets will allow you to select GSM/GPRS only.
One they really should mention because it usually catches new Mac users out: how to de-install an application. I spent some time looking for the Add/Remove Programs equivalent before I found that you just drop the application in the Trash.
"So, how did the match with the Tribbles go?"
"We do not speak about Tribbles!"
No, that's just a rounding artifact of the WP they used to write the reply. They actually spent £1e-120
(For those who don't know, Currys and PC World are branches of the same company) I'd support that: I buy a lot of stuff at PC World precisely because they've been utterly reliable on returns - they may be a bit more expensive than the web for some stuff, but it's been worth it.
On a more serious level, but for the same reason, there is no reason to think that this entry in the password file corresponds to a valid Unix password, since if that system was based on his code, he login will bypass normal authentication completely.
If you don't understand what I'm talking about - have a look at the paper. It's a classic, and well worth a read. Wikipedia has a summary .
There seem to be a lot of comments saying that this device is useless to artists because of lack of resolution and pressure sensitivity. True, but it misses the point. I used to use a Psion 7, a lovely little A5 size clamshell machine running a proprietary OS. It was ideal for taking notes in meetings as I could go straight from touch-typing to sketching a diagram straight into the word processor. These days I use MS OneNote, which is slightly more clumsy in that respect (much better in other ways). If I have a tablet I can sketch into my notes, but usually it's too much hassle to take the tablet and find room on a table in a crowded meeting room. A device that clipped on to the screen would help to a certain extent. The Psion would still be better in having the hw built in, internal stowage for the stylus, and a fancy hinge that brought the bottom of the screen forward to avoid the machine tipping backwards when the touch screen was used.
The US gallon is only 80% of the imperial gallon - it's not clear which units they are using. FWIW, I've got 47mpg (imperial) over 25000 miles in my Jaguar X-Type diesel without doing more than roughly keep to the speed limits, so I'm not overly impressed whichever units are in use.
Seems simple, fair and workable to me. There's other reasons I like them - my block of fixed public IP addresses at no extra charge, and the continuous ping-testing of my line so that they SMS me if it goes down.
Yes, we do. We're just not that impressed by it. Oh, and have you seen the size of Russia, which last time I looked was mainly an Asian country?
Just RTFA'd - that's definitely Real Physics. I'm impressed.
Update that with a modern display, WiFi, Bluetooth and HSUPA (UMTS is so two years ago) and you've got my dream machine. As to software: I've become very fond of M$ OneNote - I'd like something like that on it.
Sounds like a misunderstanding. IPv6 addresses are hierarchical. A /32 would be allocated to an ISP, and you should get a /48 from them (yes, I've done this). If your upstream ISP doesn't distribute IPv6 addresses, they aren't going to be able to route IPv6 either, so you need to find a tunnel broker. Any tunnel broker will give you a range, either a /48 or a /64, which you can use with a fixed tunnel. Alternatively you can set up a 6to4 tunnel using the anycast addresses 192.88.99.1 and 2002:c058:6301:: as the far end, which will give you a unique /48 based on the IPv4 address of your router - however if you're serious about building a commercial network on IPv6, you should probably go for the fixed link.
I've been through Kiruna a few times to go walking. It's a big, sparsely populated mining and forestry town in the Saami (Lapp) area of the north of Sweden. The air connection has to be subsidised by the government, and it's a long flight from Stockholm Arlanda in a very small plane. The air crew come round to ask who would like a taxi called for them at the airport. When you arrive, there's a single small luggage carousel and a large stuffed bear in Arrivals.
Since common law rests on precedent, old judgements are often quoted verbatim. One I particularly liked contained the phrase "every cyclist is entitled to his wobble".
I think you misunderstand what these devices are doing. They are proxies for the edge of a corporate network, and they assume that the IT manager explicitly configures the proxy settings for his users' web browsers to use the proxy (and blocks any direct Internet access using the firewall). There is an SSL/TLS connection from the browser to the proxy, and from the proxy to the target web site. They do act as a man in the middle, but this isn't in any way remarkable as it can only happen with an explicit request from the web browser.
Probably not. They burn something closer to axle grease than to oil, but it's pre-heated before injection. I've seen a similar five-cylinder engine, which used two turbochargers of 3 foot diameter, which would tend to raise the air temperature despite the use of intercoolers. In addition to those points, with cylinders that size the wall temperature isn't going to affect temperature through most of the volume of the combustion chamber simply because most of the gas isn't in contact with the wall.
It would potentially allow a network/fw admin to set policies such as "Internet Explorer may only be used to browse intranet sites"; "These versions of Firefox may be used to browse externally".
In general, you can't even send data (other than SMS) simultaneously with voice. The standards allow it, but it isn't usually implemented. Not sure how true this is for 3G phones, though.