First off, as far as I can tell the Globe is dead wrong by the second paragraph. It doesn't matter whether I photocopy, photograph or hand-copy a book, it's copyright infringement any way you look at it. So if someone uses a scanner, how is that not copyright infringement? When the Globe says someone can legally scan a book, convert it to HTML and publish it, I think they're dead wrong.
Secondly, check me but isn't what they're trying to pass law on already covered under the existing "compilation copyright" section of the copyright laws? If it is, then why should a compilation on a computer be treated any differently than, say, a phone book?
The upshot? My (uninformed) prediction is this: There will still be 40-bit non-escrowed versions of the product going out the door. These will be shipped primarily to other countries and to paranoid individuals like slashdotters. Everyone else will run 128, but it will be a compromised breed of 128.
More likely, the rest of the world and the paranoid Slashdotters will use products developed outside the US, or products like Mozilla where we can bolt whatever crypto we want into the source and chuck any escrow that tries to creep in. The politicians seem to think the whole matter is a question of they can put the holes in they want. It isn't.
That's odd, right now I just go to a search engine or directory (Yahoo, Google, et.al.), type in "clue" and get a similar list. The filtering isn't good, but that's a matter of improving the back-end implementation, not developing a new concept.
This largely makes sense. I don't know about statement 1, I don't think there *is* any absolute protection of a name.
That's why I qualified it the way I did. There is no absolute protection true, for example Apple Computers does not infringe on Apple Record's trademark of the word "Apple". You can, though, determine whether a use is not protected, eg. if I opened up Apple Computronics and advertised as Apple, Apple Computers would have a right to make me stop. IMHO domain names should be subject to the same rules as any name when it comes to trademarks, no less and no more.
2) I don't agree. "A" can have as many domains as he wants....
However, by taking up all those names, A is preventing others from using names they may be legitimately entitled to use. That's why DNS is hierarchical, to deal with this problem. IMHO if Apple wants sites for Macintosh and PowerBook computers, they should register apple.com and create macintosh.apple.com and powerbook.apple.com 3LDs, rather than registering both macintosh.com and powerbook.com. The intent of #2 is to encourage people to use 3LDs for different parts of the same entity, freeing up 2LDs for other uses.
Squatting, IMHO, can be solved by a simple rule: any offer by the holder to transfer the name for money or other valuable consideration is prima facie evidence that the name is being held only for sale and, if reported to the registrar, grounds for immediate and non-appealable revocation of the name registration. The evidence must be in the holder's own words, not paraphrased or summarized.
Assume party A holds a name, and party B wants it. My take on a reasonable set of rules:
Can B legally prevent A from using the name independent of whether it's a domain name or not? If they do so, then A loses the domain name just as they would any other name. Otherwise,
If A owns multiple domain names, B has a legal right to use the name and does not already possess a domain name, A has to either give up the name or give up all it's other names. One domain name to an entity. Otherwise,
If A does not belong under the TLD in question based on the appropriate RFCs and B does, A loses the name and B gets it. Businesses, for example, don't get to hold on to names under.org if an organization that qualifies for.org membership wants it. Otherwise,
If A has not been actively using the name, they lose it. If they're actively using it and registered it first, they keep it.
Question: what does the browser have to do with the stability of the window manager or the X server? Netscape crashes on occasion for me, but the window manager (desktop actually, I run Gnome) and X server keep running just fine (barring blockage of the X event queue, which telnet and kill deals with just fine).
I'd add to this another point: the addition of these things is a good sign for the overall Mozilla design. There's a saying about portable software, "There is no such thing as portable software, only software which has been ported.". I think the same thing applies to software intended to have relatively arbitrary plug-in functionality added. You can't design for it and get it right without actually doing it at some point. That it's being done indicates that the glitches are worked out enough to make things work, so maybe we'll have an easier time adding protocols in the future.
Verification. Signing a message in a way that prevents forgery or alteration (without having the private key) requires strong cryptography. There are lots of reasons for wanting to insure that nobody can forge messages from you, or alter one of your messages in transit.
Privacy. If I'm sending credit-card information, my address or the like to someone, it'd be nice if J. Random Bad Guy couldn't read it.
End-system privacy. Sometimes I correspond with people who have other family members using their computer. Some of what I send them is not for the consumption of those other family members for one reason or another. Without encryption, if one of those other family members picks up the mail before the intended recipient does, they get a face-full of things they were not supposed to see. Not good.
That's not counting local-system use, like insuring that even if a cracker gets on my local system he can't read the spreadsheet containing my bank account details.
Maybe this has been discussed and discarded, but is it possible to replace the Anonymous Coward with an account that has a unique name but no other identifying information? Set it up so that you could trace the history of posts by a particular person, or tell whether two posts were by the same person, but not tell who that person was? I would think this would make it easier to track the trolls and score down their posts without interfering with people who legitimately need anonymity.
Possible precautions:
If an e-mail address is needed to mail a password to, it's deleted immediately after the mail is sent and bounced password messages are bit-bucketed on arrival.
If the account isn't logged into within 48 hours after creation, it's purged.
Such accounts cannot post until at least 24 hours after first login.
Re:what is obviously being overlooked...by you
on
Hotmail Cracked Badly
·
· Score: 1
No, Microsoft didn't start Hotmail. However, Microsoft did start the Passport integration. In the course of doing this, they modified CGI scripts and failed to think through the security implications of what they were doing. Which is par for the course for MS. End result: because of a stupid error by MS, large numbers of people had e-mail compromised. In any competent setup, this error should be caught before going into production. In most Unix shops, it would get caught. Around MS, failure to catch things like this is endemic, which is why I don't trust their products from a security standpoint. I'm just happy I don't need Hotmail to get Web-based e-mail.
But like NT (or 98) you are beta tester even if you want to run production quality.
Except that, unlike NT (or 98), no Unix admin is going to commit a production system to any new version of a kernel until it's been tested and it fixes a problem that actually exists or introduces a feature that is actually needed. Even if the new version is supposedly stable. The operating principle is "if it ain't broke, don't fix it".
There's no X UI style guide, because there's no X UI. There are style guides for toolkits like Gtk and Motif that provide a UI and all the widgets, dialog boxes and such that go with it. Whether it's followed or not depends on the developer, but there's certainly some pressure to conform to the common style for each toolkit. There's even some pressure to stick to one toolkit or another, eg. to do new programs using the GTK or KDE toolkits as opposed to Motif or Athena. OTOH, if all I want is a small window that displays a pixmap and responds to mouse clicks ( ie. xbiff ), it's nice to be able to use the relatively simple Athena toolkit and not worry about Gnome/KDE overhead for something that doesn't need it.
True, but then Swing/MFC correspond to things like Gtk or Motif. The Windows GDI interface corresponds to the Xlib/Xt level in X11. Try doing raw GDI development in Windows sometime and see how "easy" that is.
The specs say that it should count the number of seconds since 1. January 1980, and it should all fit into 32 bits.
Nope. POSIX defines it to count in seconds since 1/1/1970, and I believe that it should be an integral type. Common usage says it needs to be a signed type, to allow for dates before the beginning of time. Nothing defines the size, although 32 bits is a common minimum. The easiest is to make it equivalent to int or long, which usually means 32 bits on 32-bit architectures and 64 bits on 64-bit architectures like the Alpha.
And we can practically guarantee that there'll be 32-bit machines in use in 2038. Look at the amount of 60's-era code that's still in production use today. Things aren't replaced until they break, and maybe not even then.
Check me here, but the DNS system doesn't for the most part care about the date, does it? If I reset my name server's system time to 1/1/1980 and restart it, will anything that talks to it even realize this? If not, then worst case is the root nameservers run with weird dates until the OS bugs are fixed. OS bugs may take out a couple of the roots, but I find it hard to believe that all of the root nameservers will be inoperative due to OS-level Y2K problems, and BIND isn't going to have a problem as long as the OS is working remotely sanely.
I'd take objection to his calling the use of NP-complete problems "security through obscurity", specifically his statement that we're depending on the hardness of the problems for security and that they may not be that hard if we approach them right. From what I recall, what we're depending on in reality is the fact that NP-complete problems are harder than any other known problems, and provide an upper bound on the hardness of problems. We don't know that NP-complete problems are neccesarily hard to solve, but we do know that any other problems are easier to solve. Even if someone invents a linear-time method of solving an NP-complete problem they're still harder than any other problems, the upper bound just moved down a lot. This is a problem for cryptography, but it's not solvable without rewriting the rules of mathematics.
As for calling a secret key "security through obscurity", he's missing the point. STO canonically refers to keeping the details of the underlying algorithms and/or implementation unknown. It does not typically refer to the idea that there's some piece of information that a legitimate user possesses that an attacker does not.
3dfx doesn't own DRI. DRI is the interface between XFree86, the OpenGL modules and the hardware driver that allows near-direct access to the hardware without breaking the X server in the process. 3dfx is just making a hardware driver that will follow the DRI spec so that XFree86 can use it. The XFree86 folks "own" the DRI spec.
The whole announcement doesn't excite me much. nVidia already has half-decent open-source drivers, will be ready with DRI ones when XFree86 v4 comes out and the TNT2 chips match the latest 3dfx hardware and will keep improving while 3dfx seems to have max'd out. 3dfx has annoyed me greatly with their Glide stunts. What reason do I have to buy their hardware?
Actually, never. If a program requires libc5 instead of the default glibc2.1, it uses the libc5 libraries installed. If it needs libc4, I go dig up the libc4 libraries and drop them in. If, like some versions of Netscape, it needs a specific minor version of libc5 that isn't completely compatible with the installed version of libc5, I put in the correct library and use a shell script to make sure that Netscape and Netscape alone uses that library. If I need to compile against older versions, I install the proper header files and use the correct compiler and linker switches to make them ignore the system libraries and deal with the versions I need. In short, I've never had a problem with a program not running because of incompatible libraries except when I was too lazy to install the correct libraries.
First, a casual user's understanding of "just works" mostly means "it does what I mean" and doesn't have much to do with crashes.
If it's not doing anything because it's crashed, it's not doing what they mean. And I would say that consumers are no more going to want to have to push a reset button on their EasyPC every few days than they would tolerate having to unplug their modern VCR every few days because the firmware crashed and the electronic power button wasn't responding anymore.
You miss the point completely. The whole idea of those Easy-PCs is that nobody has to set them up.
so nobody had to program the firmware on your VCR for the cable channels it receives? I'm sorry, but every piece of programmable electronics has to be set up by somebody somewhere. Even on something as simple from the end user's point of view as a VCR had to have the frequency and modulation of those channels set up by an engineer. You certainly couldn't do it, but because they already have all you have to do is tell it which band your cable company uses and punch in the channel number. If the engineers couldn't tweak frequencies, modulation schemes and such, you'd have to do it by hand using tuner dials just like you had to do on VCRs in the late 70s.
Hey, maybe your VCR does need to be "rebooted" after recording/playing tapes for a full week in a row. How would you know if it did -- have you ever used your VCR for that period of time nonstop without turning it off? Of course you haven't. (If you have, please put your white jacket with the straps back on...)
Actually it's run continuously for about 31 months now. The unit I have doesn't have a physical power switch, just an electronic one that sends it to a low-power idle mode. The CPU is still running the whole time.
I didn't have to tweak it at all, other than to set the time and program it to record programs. The guys at the manufacturer did all the tweaking of the firmware, which was probably quite a lot, and then locked it down and shipped the results. One example: they worked out the carrier frequencies and modulation for the cable channels in all 3 different bands and programmed that in, then left me with just a 3-position toggle to select which band my cable company uses. The guys at the manufacturer probably appreciate firmware that lets them tweak right down at the individual channel level, because it lets them set it up so that I don't have to. This is as opposed to an older VCR I had that didn't have this ability in the firmware and made me set the carrier frequency for each channel by hand using what amounted to a small tuner dial. This older VCR is simpler and less tweakable internally, but a lot more of a pain for the end-user than the more complex, more internally-tweakable modern unit.
(1) It crashes very rarely: important for people who depend on their computers (NOT casual users)
Actually this is important for a casual user. They want a box that, to borrow a quote, "just works". They don't want to have to deal with the fallout from crashes all the time. Even if they don't depend on it, they simply don't want to have to wonder if it's crashed after having been left alone and untouched for a week.
(2) It's a tinkerer's dream -- if you don't like something, go and change it! Again, the casual users couldn't care less.
Again, they will care because of one thing: they don't care to tinker, but the guy who set it up for them does. If the box can be tweaked, I can sit down, tweak it to exactly what's needed, lock it down so they can't touch things and duplicate the result 200 times for 200 different people.
There is practically no headache setting up mail server with linux. Provided you start using it.
Setting it up, no. Setting it up right, yes. There's anti-spam and anti-relay settings to keep track of, security patches to apply, all the headaches of looping bounce messages due to brain-dead outside mail servers and mailing lists, dealing with your machine being hammered by some idiot who tries sending 1,573,284 copies of a 3meg file to your system, etc. ad infinauseaum. All of this for 1 user. Sorry, BTDTNTSFO. If you think that installing RedHat and switching on sendmail is all there is to running a mail server, all too soon you'll get a personal introduction to the other 95%. Me, I'll let the professionals handle it
I would argue that 'always on' internet access, via cable modem and DSL is, in fact, a direct connection to the internet. And, with that, nobody needs an "Internet Service Provider," because they already have internet service.
I can think of a lot of things that an ISP provides. One example: e-mail. Suppose I have an always-on connection to the Internet. I have two choices: set up my own mail server with all the headaches that entails, or have my ISP provide me an e-mail address and run the mail server for me and everyone else they serve. Now, I'm technically capable of running my own mail server, but most Internet users aren't.
Now, my ISP provides Unix services, so I get nice things like procmail on the mail server. I depend on this. TCI, for example, doesn't provide that sort of mail service. If I have to use TCI as an ISP because I'm using their cable modem service, I can't choose to use procmail because it's not offered. Moreover, TCI won't set up DNS for me to allow me to run my own mail server, so I can't even kludge around the problem that way.
The same thing applies to DNS, FTP, HTTP. In all cases, if they decide to do something to "improve performance" like force the service through their proxies, and I have problems with their proxies or just don't want them seeing every single thing I do, I have no way of opting out of that other than selecting another ISP and if they're the only ISP on their cable that means giving up broadband access.
Re:All well and good... but it still renders poorl
on
Mozilla M8 Released
·
· Score: 2
Oh yeah, it's fast. But half the pages on the net look awful. If this product shows the inherent limitations of the real HTML specifications, then sign me up for IE.
It's probably less of a limitation of the real HTML specs as a limitation of the people who didn't follow the real HTML specs when creating their pages. My experience is that the rendering engine is positively anal about compliance, and just plain doesn't like the kludges people have put in to make things look good on IE and NS.
Slashdot Mirror
First off, as far as I can tell the Globe is dead wrong by the second paragraph. It doesn't matter whether I photocopy, photograph or hand-copy a book, it's copyright infringement any way you look at it. So if someone uses a scanner, how is that not copyright infringement? When the Globe says someone can legally scan a book, convert it to HTML and publish it, I think they're dead wrong.
Secondly, check me but isn't what they're trying to pass law on already covered under the existing "compilation copyright" section of the copyright laws? If it is, then why should a compilation on a computer be treated any differently than, say, a phone book?
Another grab by greedy companies. Feh.
The upshot? My (uninformed) prediction is this: There will still be 40-bit non-escrowed versions of the product going out the door. These will be shipped primarily to other countries and to paranoid individuals like slashdotters. Everyone else will run 128, but it will be a compromised breed of 128.
More likely, the rest of the world and the paranoid Slashdotters will use products developed outside the US, or products like Mozilla where we can bolt whatever crypto we want into the source and chuck any escrow that tries to creep in. The politicians seem to think the whole matter is a question of they can put the holes in they want. It isn't.
That's odd, right now I just go to a search engine or directory (Yahoo, Google, et.al.), type in "clue" and get a similar list. The filtering isn't good, but that's a matter of improving the back-end implementation, not developing a new concept.
This largely makes sense. I don't know about statement 1, I don't think there *is* any absolute protection of a name.
That's why I qualified it the way I did. There is no absolute protection true, for example Apple Computers does not infringe on Apple Record's trademark of the word "Apple". You can, though, determine whether a use is not protected, eg. if I opened up Apple Computronics and advertised as Apple, Apple Computers would have a right to make me stop. IMHO domain names should be subject to the same rules as any name when it comes to trademarks, no less and no more.
2) I don't agree. "A" can have as many domains as he wants....
However, by taking up all those names, A is preventing others from using names they may be legitimately entitled to use. That's why DNS is hierarchical, to deal with this problem. IMHO if Apple wants sites for Macintosh and PowerBook computers, they should register apple.com and create macintosh.apple.com and powerbook.apple.com 3LDs, rather than registering both macintosh.com and powerbook.com. The intent of #2 is to encourage people to use 3LDs for different parts of the same entity, freeing up 2LDs for other uses.
Squatting, IMHO, can be solved by a simple rule: any offer by the holder to transfer the name for money or other valuable consideration is prima facie evidence that the name is being held only for sale and, if reported to the registrar, grounds for immediate and non-appealable revocation of the name registration. The evidence must be in the holder's own words, not paraphrased or summarized.
Assume party A holds a name, and party B wants it. My take on a reasonable set of rules:
Opinions?
Question: what does the browser have to do with the stability of the window manager or the X server? Netscape crashes on occasion for me, but the window manager (desktop actually, I run Gnome) and X server keep running just fine (barring blockage of the X event queue, which telnet and kill deals with just fine).
I'd add to this another point: the addition of these things is a good sign for the overall Mozilla design. There's a saying about portable software, "There is no such thing as portable software, only software which has been ported.". I think the same thing applies to software intended to have relatively arbitrary plug-in functionality added. You can't design for it and get it right without actually doing it at some point. That it's being done indicates that the glitches are worked out enough to make things work, so maybe we'll have an easier time adding protocols in the future.
I can think of several reasons:
That's not counting local-system use, like insuring that even if a cracker gets on my local system he can't read the spreadsheet containing my bank account details.
Maybe this has been discussed and discarded, but is it possible to replace the Anonymous Coward with an account that has a unique name but no other identifying information? Set it up so that you could trace the history of posts by a particular person, or tell whether two posts were by the same person, but not tell who that person was? I would think this would make it easier to track the trolls and score down their posts without interfering with people who legitimately need anonymity.
Possible precautions:
No, Microsoft didn't start Hotmail. However, Microsoft did start the Passport integration. In the course of doing this, they modified CGI scripts and failed to think through the security implications of what they were doing. Which is par for the course for MS. End result: because of a stupid error by MS, large numbers of people had e-mail compromised. In any competent setup, this error should be caught before going into production. In most Unix shops, it would get caught. Around MS, failure to catch things like this is endemic, which is why I don't trust their products from a security standpoint. I'm just happy I don't need Hotmail to get Web-based e-mail.
But like NT (or 98) you are beta tester even if you want to run production quality.
Except that, unlike NT (or 98), no Unix admin is going to commit a production system to any new version of a kernel until it's been tested and it fixes a problem that actually exists or introduces a feature that is actually needed. Even if the new version is supposedly stable. The operating principle is "if it ain't broke, don't fix it".
There's no X UI style guide, because there's no X UI. There are style guides for toolkits like Gtk and Motif that provide a UI and all the widgets, dialog boxes and such that go with it. Whether it's followed or not depends on the developer, but there's certainly some pressure to conform to the common style for each toolkit. There's even some pressure to stick to one toolkit or another, eg. to do new programs using the GTK or KDE toolkits as opposed to Motif or Athena. OTOH, if all I want is a small window that displays a pixmap and responds to mouse clicks ( ie. xbiff ), it's nice to be able to use the relatively simple Athena toolkit and not worry about Gnome/KDE overhead for something that doesn't need it.
True, but then Swing/MFC correspond to things like Gtk or Motif. The Windows GDI interface corresponds to the Xlib/Xt level in X11. Try doing raw GDI development in Windows sometime and see how "easy" that is.
The specs say that it should count the number of seconds since 1. January 1980, and it should all fit into 32 bits.
Nope. POSIX defines it to count in seconds since 1/1/1970, and I believe that it should be an integral type. Common usage says it needs to be a signed type, to allow for dates before the beginning of time. Nothing defines the size, although 32 bits is a common minimum. The easiest is to make it equivalent to int or long, which usually means 32 bits on 32-bit architectures and 64 bits on 64-bit architectures like the Alpha.
And we can practically guarantee that there'll be 32-bit machines in use in 2038. Look at the amount of 60's-era code that's still in production use today. Things aren't replaced until they break, and maybe not even then.
Check me here, but the DNS system doesn't for the most part care about the date, does it? If I reset my name server's system time to 1/1/1980 and restart it, will anything that talks to it even realize this? If not, then worst case is the root nameservers run with weird dates until the OS bugs are fixed. OS bugs may take out a couple of the roots, but I find it hard to believe that all of the root nameservers will be inoperative due to OS-level Y2K problems, and BIND isn't going to have a problem as long as the OS is working remotely sanely.
I'd take objection to his calling the use of NP-complete problems "security through obscurity", specifically his statement that we're depending on the hardness of the problems for security and that they may not be that hard if we approach them right. From what I recall, what we're depending on in reality is the fact that NP-complete problems are harder than any other known problems, and provide an upper bound on the hardness of problems. We don't know that NP-complete problems are neccesarily hard to solve, but we do know that any other problems are easier to solve. Even if someone invents a linear-time method of solving an NP-complete problem they're still harder than any other problems, the upper bound just moved down a lot. This is a problem for cryptography, but it's not solvable without rewriting the rules of mathematics.
As for calling a secret key "security through obscurity", he's missing the point. STO canonically refers to keeping the details of the underlying algorithms and/or implementation unknown. It does not typically refer to the idea that there's some piece of information that a legitimate user possesses that an attacker does not.
3dfx doesn't own DRI. DRI is the interface between XFree86, the OpenGL modules and the hardware driver that allows near-direct access to the hardware without breaking the X server in the process. 3dfx is just making a hardware driver that will follow the DRI spec so that XFree86 can use it. The XFree86 folks "own" the DRI spec.
The whole announcement doesn't excite me much. nVidia already has half-decent open-source drivers, will be ready with DRI ones when XFree86 v4 comes out and the TNT2 chips match the latest 3dfx hardware and will keep improving while 3dfx seems to have max'd out. 3dfx has annoyed me greatly with their Glide stunts. What reason do I have to buy their hardware?
Actually, never. If a program requires libc5 instead of the default glibc2.1, it uses the libc5 libraries installed. If it needs libc4, I go dig up the libc4 libraries and drop them in. If, like some versions of Netscape, it needs a specific minor version of libc5 that isn't completely compatible with the installed version of libc5, I put in the correct library and use a shell script to make sure that Netscape and Netscape alone uses that library. If I need to compile against older versions, I install the proper header files and use the correct compiler and linker switches to make them ignore the system libraries and deal with the versions I need. In short, I've never had a problem with a program not running because of incompatible libraries except when I was too lazy to install the correct libraries.
First, a casual user's understanding of "just works" mostly means "it does what I mean" and doesn't have much to do with crashes.
If it's not doing anything because it's crashed, it's not doing what they mean. And I would say that consumers are no more going to want to have to push a reset button on their EasyPC every few days than they would tolerate having to unplug their modern VCR every few days because the firmware crashed and the electronic power button wasn't responding anymore.
You miss the point completely. The whole idea of those Easy-PCs is that nobody has to set them up.
so nobody had to program the firmware on your VCR for the cable channels it receives? I'm sorry, but every piece of programmable electronics has to be set up by somebody somewhere. Even on something as simple from the end user's point of view as a VCR had to have the frequency and modulation of those channels set up by an engineer. You certainly couldn't do it, but because they already have all you have to do is tell it which band your cable company uses and punch in the channel number. If the engineers couldn't tweak frequencies, modulation schemes and such, you'd have to do it by hand using tuner dials just like you had to do on VCRs in the late 70s.
Hey, maybe your VCR does need to be "rebooted" after recording/playing tapes for a full week in a row. How would you know if it did -- have you ever used your VCR for that period of time nonstop without turning it off? Of course you haven't. (If you have, please put your white jacket with the straps back on...)
Actually it's run continuously for about 31 months now. The unit I have doesn't have a physical power switch, just an electronic one that sends it to a low-power idle mode. The CPU is still running the whole time.
I didn't have to tweak it at all, other than to set the time and program it to record programs. The guys at the manufacturer did all the tweaking of the firmware, which was probably quite a lot, and then locked it down and shipped the results. One example: they worked out the carrier frequencies and modulation for the cable channels in all 3 different bands and programmed that in, then left me with just a 3-position toggle to select which band my cable company uses. The guys at the manufacturer probably appreciate firmware that lets them tweak right down at the individual channel level, because it lets them set it up so that I don't have to. This is as opposed to an older VCR I had that didn't have this ability in the firmware and made me set the carrier frequency for each channel by hand using what amounted to a small tuner dial. This older VCR is simpler and less tweakable internally, but a lot more of a pain for the end-user than the more complex, more internally-tweakable modern unit.
(1) It crashes very rarely: important for people who depend on their computers (NOT casual users)
Actually this is important for a casual user. They want a box that, to borrow a quote, "just works". They don't want to have to deal with the fallout from crashes all the time. Even if they don't depend on it, they simply don't want to have to wonder if it's crashed after having been left alone and untouched for a week.
(2) It's a tinkerer's dream -- if you don't like something, go and change it! Again, the casual users couldn't care less.
Again, they will care because of one thing: they don't care to tinker, but the guy who set it up for them does. If the box can be tweaked, I can sit down, tweak it to exactly what's needed, lock it down so they can't touch things and duplicate the result 200 times for 200 different people.
There is practically no headache setting up mail server with linux. Provided you start using it.
Setting it up, no. Setting it up right, yes. There's anti-spam and anti-relay settings to keep track of, security patches to apply, all the headaches of looping bounce messages due to brain-dead outside mail servers and mailing lists, dealing with your machine being hammered by some idiot who tries sending 1,573,284 copies of a 3meg file to your system, etc. ad infinauseaum. All of this for 1 user. Sorry, BTDTNTSFO. If you think that installing RedHat and switching on sendmail is all there is to running a mail server, all too soon you'll get a personal introduction to the other 95%. Me, I'll let the professionals handle it
I would argue that 'always on' internet access, via cable modem and DSL is, in fact, a direct connection to the internet. And, with that, nobody needs an "Internet Service Provider," because they already have internet service.
I can think of a lot of things that an ISP provides. One example: e-mail. Suppose I have an always-on connection to the Internet. I have two choices: set up my own mail server with all the headaches that entails, or have my ISP provide me an e-mail address and run the mail server for me and everyone else they serve. Now, I'm technically capable of running my own mail server, but most Internet users aren't.
Now, my ISP provides Unix services, so I get nice things like procmail on the mail server. I depend on this. TCI, for example, doesn't provide that sort of mail service. If I have to use TCI as an ISP because I'm using their cable modem service, I can't choose to use procmail because it's not offered. Moreover, TCI won't set up DNS for me to allow me to run my own mail server, so I can't even kludge around the problem that way.
The same thing applies to DNS, FTP, HTTP. In all cases, if they decide to do something to "improve performance" like force the service through their proxies, and I have problems with their proxies or just don't want them seeing every single thing I do, I have no way of opting out of that other than selecting another ISP and if they're the only ISP on their cable that means giving up broadband access.
Oh yeah, it's fast. But half the pages on the net look awful. If this product shows the inherent limitations of the real HTML specifications, then sign me up for IE.
It's probably less of a limitation of the real HTML specs as a limitation of the people who didn't follow the real HTML specs when creating their pages. My experience is that the rendering engine is positively anal about compliance, and just plain doesn't like the kludges people have put in to make things look good on IE and NS.