Or you could just stick insane amounts of RAM in your system, and get almost everything cached. Or perhaps not so insane these days - one gigabyte should be more than enough to have a disk cache of every file used for typical desktop / end-user tasks. Then you can set your disk to spin down after an hour of inactivity, and leave your box running for a long period. After a while most things should be cached.
The problem with this is disk writes. The disk must spin up to write data - and that includes the last access time on each file read. A write-back cache help, but not by any significant amount, and journalling makes it worse since every write must be committed to the journal immediately (ie, before the system call returns).
What I'd like to see is a way to put the journal on a separate disk, and update the main disk only infrequently. I have a 120 megabyte 2.5inch disk lying around, which should be nice and quiet. I'd like to tell ext3 to use this smaller drive as the journal file for my main disk, and importantly, only to flush the journal to disk when it really has to. So in typical use your big disk could spin down, and then the 2.5inch device would fill up with maybe a hundred megs of disk writes before the large disk needed to be spun up again for them to be committed. This would be hopeless for most servers, of course, but on a desktop it might work.
Never mind the Mac version, where's the Linux version? The list of included software sounded suspiciously Windowsy. Will there be free drivers for this card?
There should be a Slashdot poll in the manner of those after-sales questionnaires: 'how did you find out about Slashdot?'.
Myself, I started reading after Netscape's announcement that Navigator would be free software (five years later, and they still haven't shipped a version that runs fast enough to use:-P). That was following a link from the Wine newsgroup.
I'd be particularly interested in the split between those who were inducted into Slashdot by their friends or colleagues ('JOIN USSS...') and those who just found the site browsing the web. And of that latter group, how many followed explicit links and how many came from search engines?
Actually no - you can use GPLed software such as gcc for proprietary software development, and very many people do. I think the intent of the 'fields of endeavour' clause is to have no limit on _using_ the software; it is allowed to have restrictions on _distributing_ copies of the software provided they are only those needed to make sure derived works are free. (The BSD folk disagree with that last allowance.)
Hmm, 'public performance', I hadn't thought of that analogy. I wonder if the courts will accept it? Probably not, since when you publicly perform a piece of music, the audience hears the music itself. But setting up a website running some software does not distribute parts of that software - neither the source code nor the binaries. So it's hard to say that copyright is infringed.
Good point about modification being the exclusive right of the copyright holder, I can accept that even if not the 'public performance' analogy.
If the GPL version 3 does end up restricting webpageification, then I guess the FSF will have to reconsider its definition of free software a little. At the moment it is 'use, share, change' and if you have those rights then you can 'change' the software to turn it into a web page and then 'use' it. Those rights would no longer be absolute.
But that 'external deployment' stuff restricts how you may *use* the software, not just how you may distribute copies. It seems like enough to make the program non-free. Consider the DFSG: 'no discrimination against fields of endeavour'. Doesn't that mean no special requirements for using the program in a web page as opposed to a command-line app?
Also this means that the licence becomes an EULA, because it purports to restrict use of the software and not just to grant permissions for copying. The GPL's enforceability is based on copyright, but copyright (in most countries) does not require you to get permission before merely running a computer program.
Look at the text: 'you agree that any external deployment shall be deemed a distribution'. But you can't 'agree' that unless the licence is considered some kind of contract. With the GPL, it is up to copyright law to decide what counts as distributing the software. Here the licence attempts to extend copyright to count all sorts of random things as infringement, but I don't see how a court would agree with that.
Personally I've long since given up taking notice of anything the Open Source Initiative certifies. Ever since they gave their stamp of approval to that Apple licence which allows 'revokation' at any point in the future when Apple's lawyers decide not to contest a patent infringement in court. The FSF may wrap its pronouncements in ideological justification which is offputting to some, but at least when they say that a program is free software you can be sure it is.
Isn't there some more readable way to generate an IRC log? Like joining consecutive utterances by the same person into a single entry, or colour-coding to make it clearer who said what. Maybe it's just because I'm not an experienced IRC user, but these text logfiles seem almost unreadable.
There's also the point that the exclusive rights are secured *to the authors* for a limited time. Now that does not preclude the author from selling this right to a publisher. But the extension of copyright does not grant anything to the author, who is dead. Perhaps the descendents of the author could sue the publisher, saying that when the author sold his copyright interest he received payment for the 50 years (or whatever) outstanding, but never received any payment for these extra years which Congress is creating. Instead of the 20 years accruing to Disney or whoever, they should be granted to the authors' descendants.
(If copyright extension is to be granted at all, which I'd hope not - giving extra monopolies to descendents of long-dead authors is a sucky idea and not really any better than subsidizing Disney.)
Actually I was just comparing with my InfoMagic Linux CDs from 1995 and 1996. They are so heavy and chunky compared with later discs. Also they don't bend alarmingly when you try to pull them out of the holder.
We're all familiar with describing someone as a 'smoothie'. A person with a certain amount of charm, or who at least thinks he has charisma. A related term is 'smooth operator'. Such people might typically be working in marketing or public relations, or perhaps as politicians or a certain type of PHB.
It doesn't take much thought to realize that the word 'smooth' comes from an association with being clean-shaven. After all, it is applied only to men. If you want more evidence, how about the epithet 'smoothychops'.
Now consider a theoretician or a Real Programmer. Surely the first image that comes to mind is the possibly-overweight and heavily bearded man in loosely-fitting clothes. In a typical technology company, these people are at the opposite end from the marketroids; but despite their strong technical knowledge they may not always be able to apply it practically (to the end of making money, at least).
So we have at one end the clean-shaven, 'smooth' but superficial and essentially useless marketing half of a technology company. At the other, the -bearded but also somewhat unrealistic technical side. But in the programme 'Rough Science', competitors are expected to have theoretical knowledge and also to apply it successfully. The title refers to the several days' stubbly beard growth a typical male scientist will get after a few days stranded on the island. This 'rough', newly-grown beard is a blend of the two facial hair types.
Yeah, the arguments against firewalls are very similar to those against VPNs.
I don't have a problem with using these as an additional layer of defence - as well as secure, encrypted protocols - but too often they seem to be slapped on top of an existing insecure setup rather than fixing the underlying problems.
But you and most people here will already know all that.
I was thinking of CD boxed sets (especially places like Cheap Bytes). Any reduction in the price and shipping costs is worthwhile. I was pissed off that my Mandrake CDs didn't come in a nice holder, but maybe they had an excuse in that holders for eight or so CDs are hard to get hold of. It would have been handier and less bulky to ship four double-sided CDs.
For your own burning... maybe you wouldn't want different things on both sides because of the labelling difficulties. But still, you could use part of the disk - say a 1cm band around the edge on one side - for labelling, and still have well over a gigabyte of storage per disk.
IMO, anything that could have reduced the number of CDs currently stored in drawers and boxes is a good thing. And unlike switching to DVD, it doesn't require any new hardware (unless you count the five fingers used to pick up the disk and turn it over).
Maybe you could also label the damn things just by having 'dead patches' on the CD where the ink goes. One byte of text on the label costs you 500 kilobytes in lost storage capacity:-).
But CDs are much thinner now than they used to be... flimsy little things, not chunky like in the good old days. So if they doubled the thickness it'd still be within the CD specifications, I expect.
I meant that hardware would boot a protected-mode OS and run Windows under virtualization, as opposed to the current setup when the bootloader uses ye olde real mode and then Windows takes over the whole machine. The point is that there would be a layer between Windows and the hardware.
If virtualization software becomes cheap and effective, perhaps PC clones five years from now will always boot into Linux or another protected-mode OS and then start Windows on the virtual hardware. It would be a nice way to add more power to BIOS setup screens or to allow installing multiple OSes with no hassle. Perhaps also to let Windows 1984 run even on hardware that hasn't been certified as 'trusted'.
You make a good point - 'show me a secure protocol that allows you to mount your home directory across a network'. There isn't one, or if there is then it's not widely deployed and doesn't have the maturity of NFS. However, if the protocol is insecure, then doesn't VPNing it just hide the problem? It does nothing to protect you from internal attacks.
I accept that hiding the problem from the outside world is better than leaving it exposed, but still the best answer would be to fix the problem altogether (using Kerberized NFS, if such a thing existed, or just ssh/scp).
I think TV text reading depends on the particular font and not just enlarging ordinary fonts. With a bitmapped font at say 80x30 screen size it should be usable.
What's needed is for somebody to go through and rip out all the effete X11 bitmap fonts, replacing them with nice chunky typefaces which have bitmaps for 80 columns on a TV screen (so one em would be about ten pixels wide). One of the tactics used by some 'home computers' was to make each vertical line two pixels wide rather than just one.
Stuff like wordprocessing is harder to do; even with really good anti-aliasing you're unlikely to get a pleasant WYSIWYG display. Again, you should rely on ready-made bitmaps for common sizes (as already happens sometimes) but these bitmaps could be anti-aliased and optimized for the horizontal blurring caused by TV screens.
If you use a VPN, you probably don't care much about security anyway. VPNs are useful only when you have servers which grant access based on source IP address or other such nonsense. If you used secure protocols to start with (ssh, https with authentication) then you wouldn't need the VPN.
OK - this is a troll - but could someone explain whether VPNs have any real uses apart from working around insecure servers which trust the network too much. QoS is one thing perhaps, but it seems like overkill for that.
If you need to wrap clear-text protocols with an encryption layer, isn't ssh tunnelling a better and simpler solution?
They could send you a computer with one month's free AOL, and after that time period come to collect it back.
Or you might rent a PC as part of your AOL subscription - for $6 / month extra this is certainly possible, if the hardware costs $150 to manufacture and lasts a couple of years. It might even save AOL money by reducing support costs.
All that's needed is some way of getting a usable display on a TV screen... (I'm not optimistic).
The PPro and Athlon are _hardware_ x86 emulators (as is every x86-compatible chip, in a way). I was referring to software emulation, where a program is loaded into memory and run to provide a simulated x86 chip.
You may be right that code density is a reason to keep x86 machine code, but clearly it's not the only reason. If you were asked to design a bytecode from scratch and to optimize code density, you wouldn't end up with 80386 assembler:-P. Established base of binary-only programs must be the main reason.
Yeah but so far every platform that has tried to emulate x86 processors in software has dismally failed to make inroads into the PC market. Even in cases where the software emulation ran _faster_ than the equivalently priced PC it hasn't helped.
Slashdot Mirror
Or you could just stick insane amounts of RAM in your system, and get almost everything cached. Or perhaps not so insane these days - one gigabyte should be more than enough to have a disk cache of every file used for typical desktop / end-user tasks. Then you can set your disk to spin down after an hour of inactivity, and leave your box running for a long period. After a while most things should be cached.
The problem with this is disk writes. The disk must spin up to write data - and that includes the last access time on each file read. A write-back cache help, but not by any significant amount, and journalling makes it worse since every write must be committed to the journal immediately (ie, before the system call returns).
What I'd like to see is a way to put the journal on a separate disk, and update the main disk only infrequently. I have a 120 megabyte 2.5inch disk lying around, which should be nice and quiet. I'd like to tell ext3 to use this smaller drive as the journal file for my main disk, and importantly, only to flush the journal to disk when it really has to. So in typical use your big disk could spin down, and then the 2.5inch device would fill up with maybe a hundred megs of disk writes before the large disk needed to be spun up again for them to be committed. This would be hopeless for most servers, of course, but on a desktop it might work.
Never mind the Mac version, where's the Linux version? The list of included software sounded suspiciously Windowsy. Will there be free drivers for this card?
BBC News online has a fine tradition of banal or inappropriate picture 'bylines'. NTK has chronicled many, such as: 0, 1, 2.
There should be a Slashdot poll in the manner of those after-sales questionnaires: 'how did you find out about Slashdot?'.
:-P). That was following a link from the Wine newsgroup.
Myself, I started reading after Netscape's announcement that Navigator would be free software (five years later, and they still haven't shipped a version that runs fast enough to use
I'd be particularly interested in the split between those who were inducted into Slashdot by their friends or colleagues ('JOIN USSS...') and those who just found the site browsing the web. And of that latter group, how many followed explicit links and how many came from search engines?
Actually no - you can use GPLed software such as gcc for proprietary software development, and very many people do. I think the intent of the 'fields of endeavour' clause is to have no limit on _using_ the software; it is allowed to have restrictions on _distributing_ copies of the software provided they are only those needed to make sure derived works are free. (The BSD folk disagree with that last allowance.)
Hmm, 'public performance', I hadn't thought of that analogy. I wonder if the courts will accept it? Probably not, since when you publicly perform a piece of music, the audience hears the music itself. But setting up a website running some software does not distribute parts of that software - neither the source code nor the binaries. So it's hard to say that copyright is infringed.
Good point about modification being the exclusive right of the copyright holder, I can accept that even if not the 'public performance' analogy.
If the GPL version 3 does end up restricting webpageification, then I guess the FSF will have to reconsider its definition of free software a little. At the moment it is 'use, share, change' and if you have those rights then you can 'change' the software to turn it into a web page and then 'use' it. Those rights would no longer be absolute.
But that 'external deployment' stuff restricts how you may *use* the software, not just how you may distribute copies. It seems like enough to make the program non-free. Consider the DFSG: 'no discrimination against fields of endeavour'. Doesn't that mean no special requirements for using the program in a web page as opposed to a command-line app?
Also this means that the licence becomes an EULA, because it purports to restrict use of the software and not just to grant permissions for copying. The GPL's enforceability is based on copyright, but copyright (in most countries) does not require you to get permission before merely running a computer program.
Look at the text: 'you agree that any external deployment shall be deemed a distribution'. But you can't 'agree' that unless the licence is considered some kind of contract. With the GPL, it is up to copyright law to decide what counts as distributing the software. Here the licence attempts to extend copyright to count all sorts of random things as infringement, but I don't see how a court would agree with that.
Personally I've long since given up taking notice of anything the Open Source Initiative certifies. Ever since they gave their stamp of approval to that Apple licence which allows 'revokation' at any point in the future when Apple's lawyers decide not to contest a patent infringement in court. The FSF may wrap its pronouncements in ideological justification which is offputting to some, but at least when they say that a program is free software you can be sure it is.
Isn't there some more readable way to generate an IRC log? Like joining consecutive utterances by the same person into a single entry, or colour-coding to make it clearer who said what. Maybe it's just because I'm not an experienced IRC user, but these text logfiles seem almost unreadable.
I think they are actually thicker, laying one side by side with a new thin CD. But I'll have to check.
There's also the point that the exclusive rights are secured *to the authors* for a limited time. Now that does not preclude the author from selling this right to a publisher. But the extension of copyright does not grant anything to the author, who is dead. Perhaps the descendents of the author could sue the publisher, saying that when the author sold his copyright interest he received payment for the 50 years (or whatever) outstanding, but never received any payment for these extra years which Congress is creating. Instead of the 20 years accruing to Disney or whoever, they should be granted to the authors' descendants.
(If copyright extension is to be granted at all, which I'd hope not - giving extra monopolies to descendents of long-dead authors is a sucky idea and not really any better than subsidizing Disney.)
Actually I was just comparing with my InfoMagic Linux CDs from 1995 and 1996. They are so heavy and chunky compared with later discs. Also they don't bend alarmingly when you try to pull them out of the holder.
We're all familiar with describing someone as a 'smoothie'. A person with a certain amount of charm, or who at least thinks he has charisma. A related term is 'smooth operator'. Such people might typically be working in marketing or public relations, or perhaps as politicians or a certain type of PHB.
It doesn't take much thought to realize that the word 'smooth' comes from an association with being clean-shaven. After all, it is applied only to men. If you want more evidence, how about the epithet 'smoothychops'.
Now consider a theoretician or a Real Programmer. Surely the first image that comes to mind is the possibly-overweight and heavily bearded man in loosely-fitting clothes. In a typical technology company, these people are at the opposite end from the marketroids; but despite their strong technical knowledge they may not always be able to apply it practically (to the end of making money, at least).
So we have at one end the clean-shaven, 'smooth' but superficial and essentially useless marketing half of a technology company. At the other, the -bearded but also somewhat unrealistic technical side. But in the programme 'Rough Science', competitors are expected to have theoretical knowledge and also to apply it successfully. The title refers to the several days' stubbly beard growth a typical male scientist will get after a few days stranded on the island. This 'rough', newly-grown beard is a blend of the two facial hair types.
Yeah, the arguments against firewalls are very similar to those against VPNs.
I don't have a problem with using these as an additional layer of defence - as well as secure, encrypted protocols - but too often they seem to be slapped on top of an existing insecure setup rather than fixing the underlying problems.
But you and most people here will already know all that.
I was thinking of CD boxed sets (especially places like Cheap Bytes). Any reduction in the price and shipping costs is worthwhile. I was pissed off that my Mandrake CDs didn't come in a nice holder, but maybe they had an excuse in that holders for eight or so CDs are hard to get hold of. It would have been handier and less bulky to ship four double-sided CDs.
:-).
For your own burning... maybe you wouldn't want different things on both sides because of the labelling difficulties. But still, you could use part of the disk - say a 1cm band around the edge on one side - for labelling, and still have well over a gigabyte of storage per disk.
IMO, anything that could have reduced the number of CDs currently stored in drawers and boxes is a good thing. And unlike switching to DVD, it doesn't require any new hardware (unless you count the five fingers used to pick up the disk and turn it over).
Maybe you could also label the damn things just by having 'dead patches' on the CD where the ink goes. One byte of text on the label costs you 500 kilobytes in lost storage capacity
But CDs are much thinner now than they used to be... flimsy little things, not chunky like in the good old days. So if they doubled the thickness it'd still be within the CD specifications, I expect.
I meant that hardware would boot a protected-mode OS and run Windows under virtualization, as opposed to the current setup when the bootloader uses ye olde real mode and then Windows takes over the whole machine. The point is that there would be a layer between Windows and the hardware.
Why have a 'media side' at all? Why not have data on both sides of a CD?
If virtualization software becomes cheap and effective, perhaps PC clones five years from now will always boot into Linux or another protected-mode OS and then start Windows on the virtual hardware. It would be a nice way to add more power to BIOS setup screens or to allow installing multiple OSes with no hassle. Perhaps also to let Windows 1984 run even on hardware that hasn't been certified as 'trusted'.
You make a good point - 'show me a secure protocol that allows you to mount your home directory across a network'. There isn't one, or if there is then it's not widely deployed and doesn't have the maturity of NFS. However, if the protocol is insecure, then doesn't VPNing it just hide the problem? It does nothing to protect you from internal attacks.
I accept that hiding the problem from the outside world is better than leaving it exposed, but still the best answer would be to fix the problem altogether (using Kerberized NFS, if such a thing existed, or just ssh/scp).
I think TV text reading depends on the particular font and not just enlarging ordinary fonts. With a bitmapped font at say 80x30 screen size it should be usable.
What's needed is for somebody to go through and rip out all the effete X11 bitmap fonts, replacing them with nice chunky typefaces which have bitmaps for 80 columns on a TV screen (so one em would be about ten pixels wide). One of the tactics used by some 'home computers' was to make each vertical line two pixels wide rather than just one.
Stuff like wordprocessing is harder to do; even with really good anti-aliasing you're unlikely to get a pleasant WYSIWYG display. Again, you should rely on ready-made bitmaps for common sizes (as already happens sometimes) but these bitmaps could be anti-aliased and optimized for the horizontal blurring caused by TV screens.
If you use a VPN, you probably don't care much about security anyway. VPNs are useful only when you have servers which grant access based on source IP address or other such nonsense. If you used secure protocols to start with (ssh, https with authentication) then you wouldn't need the VPN.
OK - this is a troll - but could someone explain whether VPNs have any real uses apart from working around insecure servers which trust the network too much. QoS is one thing perhaps, but it seems like overkill for that.
If you need to wrap clear-text protocols with an encryption layer, isn't ssh tunnelling a better and simpler solution?
They could send you a computer with one month's free AOL, and after that time period come to collect it back.
Or you might rent a PC as part of your AOL subscription - for $6 / month extra this is certainly possible, if the hardware costs $150 to manufacture and lasts a couple of years. It might even save AOL money by reducing support costs.
All that's needed is some way of getting a usable display on a TV screen... (I'm not optimistic).
The PPro and Athlon are _hardware_ x86 emulators (as is every x86-compatible chip, in a way). I was referring to software emulation, where a program is loaded into memory and run to provide a simulated x86 chip.
:-P. Established base of binary-only programs must be the main reason.
You may be right that code density is a reason to keep x86 machine code, but clearly it's not the only reason. If you were asked to design a bytecode from scratch and to optimize code density, you wouldn't end up with 80386 assembler
Yeah but so far every platform that has tried to emulate x86 processors in software has dismally failed to make inroads into the PC market. Even in cases where the software emulation ran _faster_ than the equivalently priced PC it hasn't helped.
What? Intel are officially admitting that overclocking works, and that hardware sold with a rated speed of X GHz may actually be capable of much more?