But if X is both ambiguous and unimportant, you're just communicating useless information.
True. But this situation is rare. Note that few people suggest *never* using the passive; that would be stupid. But many people use the passive in situations where it is inappropriate, where information about the agent of the actions being described is known, or where it would be useful for the reader to know that the agent is unknown. Passive sentences are also often confusing structurally, making documents harder to read. These arguments have convinced me that it is better to use the active wherever possible -- that is, in situations that are not as you describe, where the actor is unknown and unimportant.
But just by adding more information doesn't necessarily make the passive more justified than without that information.
True, but if you can add the extra information (identity of the actor) that is required to make the sentence active, then I find the sentence is usually better phrased in the active voice. Passive sentences beyond a certain level of complexity are frequently clumsy and difficult to read. It is often far easier to understand an active version of the same sentence. This is why so many stylists recommend using the active whenever possible -- you may be able to parse your passive sentences just fine, so you might not realise the difficulty they cause for your readers.
I've played around with the LinkGrammar parser that they've based it on before, and it's pretty good. I'd say it is capable of parsing a lot of sentences that MS Word's grammar checker gets hopelessly confused by. Unfortunately, it's quite slow, as it apparently has to search a tree that is exponential on the number of words in the sentence (I think). I certainly wouldn't expect to be able to use it for real-time highlighting like MS Word supports.
I'd be more concerned that if it were GPL'd that it couldn't use some or all of the above.
Anything that can be distributed as LGPL can also be distributed as GPL if required -- that's actually a requirement of the LGPL that is necessary to ensure that the LGPL is a GPL-compatible license (otherwise GPL programs couldn't use LGPL libs, which would be ridiculous). So there'd clearly be nothing to worry about.
You are of course perfectly free to make sonamchauhanoffice, incorporating code from openoffice.org and linkgrammar.
However, because Sun bases its proprietary StarOffice on openoffice, code where the copyright can't be assigned to sun for relicensing is unlikely to make it into their repository.
Huh? LinkGrammar is distributed under a BSD-like license. This means that Sun are legally allowed to use it, as long as they follow certain very simple restrictions (as they already do for other libraries that they depend on, e.g. libcurl). Copyright assignment has nothing to do with it; they certainly haven't had the copyright for libcurl, aspell, berkely DB, freetype, etc. assigned to them, and nor is there any reason they should.
See here for a list of components in OpenOffice that were not written by OO.o developers. There are more in StarOffice that couldn't be released to OO.o because Sun couldn't get permission from the copyright holders of those components to license them appropriately (e.g. the database system).
Go get a clue. The suggestion that AbiWord is able to do this because it is GPL is a blatant troll, and doesn't merit defending.
I'm starting to find the pro-GPL tone of many developers very annoying, actually. Some of us don't license our code under the GPL because we don't agree with the philosophy of it -- this doesn't mean we want to have the "advantages" of the GPL rammed down our throats at every opportunity.
It's simple: a very large proportion of libraries are available either under the LGPL, or (like the one used by AbiWord here) a BSD-like license. Reusing code from projects that are not libraries is often hard work, and it is frequently easier to either find a library that does the job, or start from scratch. This means that for almost all purposes the supposed advantage of being able to reuse GPL code is actually useless.
Punitive damages are awarded only in special cases where conduct was egregiously invidious
From WordNet (via dictionary.com):
invidious adj : containing or implying a slight or showing prejudice; "discriminatory attitudes and practices"; "invidious comparisons"
I don't think this is a description that can really be applied to downloading copyright files without the holder's consent, so punitive damages should not be awarded in these cases (assuming the wikipedia text is correct and complete).
Hardware-support is a no-brainer. It's really simple: *do your research before you buy*, and it will be equally well supported in Linux.
Yeah, right. We live in a world where vendors change the chipsets in their cards without changing the model numbers. No amount of research will tell you whether or not a Belkin F5D7200 will work under Linux: it could be one of two entirely different cards, one of which works fine, the other of which doesn't, and you won't be able to tell until you get it out of the box. And if you're not very careful with the research you do, you'll just get the "it works fine" response, because it's listed on the hardware compatibility lists.
And when that's the only card that your local shops stock that does the job, are you supposed to travel further to a different shop? Where do you draw the line at what's convenient and what isn't?
The point is, if you run Windows, you can just buy hardware, take it out of the box, plug it in, stick the CD in the drive and expect it to work. If you run Linux, about 60% of the time it's even easier. But the other 40% is a real hassle.
Can you tell me how price fixing and other anti-competitive practices did this?
Easy. By consolidating the market and forcing it into a virtual monopoly, they've been able to substantially cut their prices in real terms (the cost of DOS + Windows at the time of Windows' release was about $150, which is roughly the same as a copy of WinXP Home, so there's 20 years worth of inflation that has come off that figure). Low prices make the PC platform look more attractive than alternatives (like Macs).
While obsoleting your computer every two or three years, they also make sure no one can sell you a new one that does not put money in their pocket.
That's funny. I'm sitting here writing this using a computer that's over 6 years old, yet I'm running the latest version of Windows without much difficulty. Also, my local computer store is quite capable of selling me a computer without any MS software, thanks.
In the early days, MS helped compatible system builders substantially by building customized versions of DOS (and even early Windows versions) that would run on machines that weren't 100% BIOS/hardware compatible with IBM. By doing this, they stimulated the competition in a way that wouldn't have been possible if, say, IBM had produced their own OS for the PC.
Although I still wonder where we'd be right now if IBM had bought CP/M 86 instead.
I hope you do realize that there's a difference between "spyware", "virus" and "worm". Hint: "spyware" is usually installed with the user's unknowing "consent".
Actually, I'm wondering whether you do. "Spyware" (and its annoying cousin, "adware," which randomly pops up adverts while you're trying to work) relates to what the software does when it is on your computer, whereas virus, worm, and trojan horse are all methods of distribution, and could perform any function desired once they have reached the computer (although by the definition of virus and worm the function must at least include spreading the malware further).
The spyware you're talking about is distributed in trojan horse form, but there's no reason it can't be distributed as a virus or worm. It is also commonly distributed in a fashion which doesn't quite merit being called a worm (because it isn't capable of spreading itself after it has installed) but isn't anything else I know a name for: it sits on a web site (with the web site owners cooperation) and infects passing browsers that are vulnerable to particular attacks. This is how better security could help prevent spyware, by eliminating this installation vector.
Then there was Windows 2000, and it was more stable and still easy to use. Windows XP could hav been a Windows 2000 service pack. A better themable UI, a minor IE update, some utilities to do things like registry snapshots that were useful, but always available as cheap third party tools. No big deal.
Well, Win2K = NT 5.0 and WinXP = NT 5.1, released only a year and a half later, so what were you expecting?
That said, a lot of the useful features that were supposed to be in Win2K from the start (particularly remote desktop and fast user switching) did get shoved back to WinXP, so I wouldn't want to have stuck with 2K.
XP SP 2 was the same, except the firewall was so bad you still needed a third party firewall. And yeah, spyware got more popular in the last few years, so you need antispyware tools now too.
OK, this I don't get. What is wrong with the XP SP2 firewall? It does the job its supposed to do: stop incoming connections to services that you haven't specifically authorised to accept connections from outside. And please don't tell me it's because it doesn't do egress filtering: egress filtering is just about useless. It's trivially easy for any malware that wants to send data outbound to do so using a method that could not possibly be distinguished from legitimate traffic.
And yeah, spyware got more popular in the last few years, so you need antispyware tools now too.
Funnily enough, I've never had a problem with it. And I've never needed a tool to get rid of it off other people's systems, either. As long as you don't go around installing random software from unknown sources, you won't have a problem. Of course, Linux users don't do this, because unknown sources don't tend to have a Linux version of their software available, so it isn't really an issue there, either.
Sure, you could install all this stuff in Windows, but you have to find it and pay for it and reboot and reboot and reboot.
Everything you list is trivially easy to find, completely free and doesn't require reboots to install under Windows these days.
There's even useful shit like strace in the OS.
Similar tools are available for windows as parts of the various SDKs, or for independent download. The rationale for not including it is that it's useless to anyone who isn't a developer, which seems sensible to me. Linux installs are often intimidating to new users, because they get to the software selection screen and see all the packages that are included, and have no idea how to choose what they need and what they don't need. Windows works the other way around: give the user a basic install with some tools that a large proportion of users will find useful (but without asking what they want), and then let them get anything else afterwards. It's a difference in philosophy and there advantages each way. But what it isn't, really, is a big deal.
Scenario A: I download an open source software, version 0.7.8.1 that does 80% of what I need. The project is either not in active development 'cause the developer got bored, or he/she/they refuse to add the features I want. I then look for a freelance programmer. I can expect at least 3 months of work; even at $8/hour, that's still $3840 for someone who works 40 hours per week.
Your estimate of the amount of work involved for fixing the problems you have with it is over the top. Chances are that your v0.7.8.1 program hasn't had the 480 hours of time that you reckon you'll need put into it at all, and that's probably between 2 or 3 programmers -- your single programmer with a fixed goal will be able to work more efficiently, because he doesn't need to communicate with other programmers to find out how they think things should work, integrate with modules they've written while he's been working on something else, etc: he just has to work to your spec.
Case in point: for years we've been using at work a Linux Debian as firewall / gateway / DHCP server. Every time we needed a new feature (VPN, bandwidth management) we needed to call a specialist. The box itself was free, but each configuration cost us about $250 [...] after paying $2500 I went out and bought a Zyxell Zywall 10W with wireless for $200, that does 98% of what I need.
You see, the problem with what you did here is that you tried to use a non-commercial Linux distribution in a commercial environment. While I still wouldn't recommend it if the dedicated device does what you need, a better way of doing this would have been to get a commercially supported Linux distribution (e.g. SuSE Server - $350); this would come with installation support to walk you through setting up what you need. If you later need additional support (e.g. because you need to add new features) you can purchase it at $900 per annum. This is a much cheaper way of doing what you wanted.
Since then I keep asking myself: why couldn't be Linux so easy to use, with clearly organized menus for everthing?
Many versions of Linux do come with such menus for a large portion of their features. Again, you made a mistake choosing Debian: its target market is geeks.
There's no convincing them to drive on the right though.
But... if we were to drive on the right, and a highwayman were to attack us while he passed us coming the opposite way, we would have to defend ourselves with our left hands, and that's just not cricket!:)
Isn't that already the case ? If the domain isn't specified, the DNS server will check if there's a machine of that name in the current domain.
Yes, but most computers on the Internet these days don't actually know what their local DNS name is (if they have one at all, that is). That system stopped being generally useful sometime around '96 or '97.
Sequential logic is hard to understand, and as project complexity gets higher it becomes exponentially more difficult to comprehend the possible interactions and determine what will happen in any particular circumstance.
The potential for research in modular non-sequential programming systems is great: a reliable, easy-to-use toolkit along these lines would become very popular very quickly, as it would make programming a lot easier than it is right now.
Sorry, region coding and control disabling (which is what prevents you from skipping the adverts) are not the same thing at all. I think HDDVD will still have the latter. They like that feature.
True, but the point is, because you've standardised on a single supplier, you automatically know who to call when you have a dead machine.
If you've been switching suppliers frequently to get the best deal on any particular component, you'll have to look up in your records to find out who you got that particular board from. And heaven help you if two different machines have had their labels confused at some point...
Fair enough, but I think google buys a lot more computers than the OP. Also, the PCs they need are very specialised, and they probably can't get a good deal on them from a mainline supplier like Dell, who typically package components google wouldn't need with all machines (CDROM, floppy drive, keyboard, mouse).
Its not only the price for the systems that randomly changes. I was on a client site recently while the client ordered some new hardware; about 15 Dell PCs. The quote came back roughly what he was expecting for the hardware (this was the second batch of similar machines he had ordered), but the delivery charge was astronomical. He called his sales rep, who informed him that they now charge a per-PC delivery charge, not a flat one. He cancelled the order because he would have to get clearance from his boss... the next day a new quote turns up with the same delivery charge as the original batch.
I've looked at those in a little detail, and while I'm not an expert I think I'm correct in the following summary:
1. The first is a patent on the long filename support of Win95 and later versions. It covers the method used for storing the long and short filenames in the same directory and for hiding the long filenames from older versions of DOS. , along with a few simple variations.
2. The second is a broader patent which just covers the same ideas but breaks them down into a larger number of simpler claims (the second being rather amusing: "2. The method as recited in claim 1 wherein the long filename contains more characters than the short filename." Well, duh!)
3. The third is for an API that allows DOS programs to use long filenames. Many of the claims are ridiculously broad and should easily be defeated by prior art claims:
Claim 1 basically reads "any extension to MS-DOS for using 64 bit file times". This is patenting the problem, not the solution, and therefore shouldn't stand. Claim 2 is "the above, but using the same interface numbers we've used". The interface numbers chosen are an inconsequential detail, and do not satisfy any reasonable definition of non-obviousness. Claim 3 is "any operating system that implements both short and long files and can convert 64 bit file times to and from BCD." To sink this one, you need to find an OS that predates Win95 that allowed you to create both short and long filenames referring to the same file. I'd be surprised if this hadn't been done before MS did it. The date conversion features don't seem to satisfy the criterion of non-obviousness to me. Claim 4 is "any operating system that automatically creates a short name when you tell it to create a file with a long name." This might be the hardest claim to sink. They might actually have invented this feature. Claim 5 is the same, except you use "int 21h" as the syscall function (i.e. the OS is DOS or CP/M 86 compatible) Claim 6 is the same as 5, except the function number is 7139h. Claim 7 is the same as 4, except it's for deleting files, rather than creating them. Claims 8 and 9 are equivalent to 5 and 6. Claims 10-12 are a minor variation on claims 7-9. Claims 13-15 are the equivalents for renaming files Claims 16-19 are the equivalents for listing directories Claims 20-24 are the same as the above claims 4, 7, 10, 13 and 16, except that they are for disks with the software stored on them, rather than a computer with the software loaded.
It's worth noting that all of these were originally filed prior to the release of win95, but the application was abandoned and a new application filed at a later date.
Hell, that's hardly the most egregious error in the document. Apparently, the FAT format is used by Samba, the open source software that allows Linux to share files with Windows and read files from Windows disks.
I kind of agree with your argument. There are just two problems with it:
1. History has shown that people with strong leanings in any direction are very frequently unable to overcome those leanings when interpreting facts before them. Two different people can read the same facts in entirely different ways because of this.
2. If a case were as simple as reading the law and systematically applying to it, it would not reach the supreme court. Cases that get as far as the supreme court are those that require interpretation of the more complex aspects of law, especially the interactions between different laws that may have been written with different goals in mind. This interpretation can frequently be a highly personal matter. This is why many supreme court judgements are not unanimous.
Because of these facts, I think it is best if all judges have no strong leanings towards any particular political viewpoint, other than those that are actually embodied in the laws of the land.
IIRC, there wasn't really a bug in the disk compression; the issue was lazy writes. Users who, for example, shut the system down precipitously, might have had written data lost, but the same is true of any modern OS. I think all 6.2 did in this regard was to shut off lazy writes by default.
Right. The problem is, with DOS, how do you know when the data has been written to disk? Turns out the common recommendation was to wait until you got the DOS prompt back after quitting your apps, but this wasn't adequate protection. The fix was to make command.com flush the disk before displaying the prompt every time.
Slashdot Mirror
But if X is both ambiguous and unimportant, you're just communicating useless information.
True. But this situation is rare. Note that few people suggest *never* using the passive; that would be stupid. But many people use the passive in situations where it is inappropriate, where information about the agent of the actions being described is known, or where it would be useful for the reader to know that the agent is unknown. Passive sentences are also often confusing structurally, making documents harder to read. These arguments have convinced me that it is better to use the active wherever possible -- that is, in situations that are not as you describe, where the actor is unknown and unimportant.
But just by adding more information doesn't necessarily make the passive more justified than without that information.
True, but if you can add the extra information (identity of the actor) that is required to make the sentence active, then I find the sentence is usually better phrased in the active voice. Passive sentences beyond a certain level of complexity are frequently clumsy and difficult to read. It is often far easier to understand an active version of the same sentence. This is why so many stylists recommend using the active whenever possible -- you may be able to parse your passive sentences just fine, so you might not realise the difficulty they cause for your readers.
You've also ended a sentence with a preposition, which you're not supposed to do... ;-)
In the words of Winston Churchill, "This is the sort of bloody nonsense up with which I will not put."
...have a GOOD grammar checker?
I've played around with the LinkGrammar parser that they've based it on before, and it's pretty good. I'd say it is capable of parsing a lot of sentences that MS Word's grammar checker gets hopelessly confused by. Unfortunately, it's quite slow, as it apparently has to search a tree that is exponential on the number of words in the sentence (I think). I certainly wouldn't expect to be able to use it for real-time highlighting like MS Word supports.
I'd be more concerned that if it were GPL'd that it couldn't use some or all of the above.
Anything that can be distributed as LGPL can also be distributed as GPL if required -- that's actually a requirement of the LGPL that is necessary to ensure that the LGPL is a GPL-compatible license (otherwise GPL programs couldn't use LGPL libs, which would be ridiculous). So there'd clearly be nothing to worry about.
You are of course perfectly free to make sonamchauhanoffice, incorporating code from openoffice.org and linkgrammar.
However, because Sun bases its proprietary StarOffice on openoffice, code where the copyright can't be assigned to sun for relicensing is unlikely to make it into their repository.
Huh? LinkGrammar is distributed under a BSD-like license. This means that Sun are legally allowed to use it, as long as they follow certain very simple restrictions (as they already do for other libraries that they depend on, e.g. libcurl). Copyright assignment has nothing to do with it; they certainly haven't had the copyright for libcurl, aspell, berkely DB, freetype, etc. assigned to them, and nor is there any reason they should.
See here for a list of components in OpenOffice that were not written by OO.o developers. There are more in StarOffice that couldn't be released to OO.o because Sun couldn't get permission from the copyright holders of those components to license them appropriately (e.g. the database system).
Go get a clue. The suggestion that AbiWord is able to do this because it is GPL is a blatant troll, and doesn't merit defending.
I'm starting to find the pro-GPL tone of many developers very annoying, actually. Some of us don't license our code under the GPL because we don't agree with the philosophy of it -- this doesn't mean we want to have the "advantages" of the GPL rammed down our throats at every opportunity.
It's simple: a very large proportion of libraries are available either under the LGPL, or (like the one used by AbiWord here) a BSD-like license. Reusing code from projects that are not libraries is often hard work, and it is frequently easier to either find a library that does the job, or start from scratch. This means that for almost all purposes the supposed advantage of being able to reuse GPL code is actually useless.
From the article you reference:
Punitive damages are awarded only in special cases where conduct was egregiously invidious
From WordNet (via dictionary.com):
invidious adj : containing or implying a slight or showing prejudice; "discriminatory attitudes and practices"; "invidious comparisons"
I don't think this is a description that can really be applied to downloading copyright files without the holder's consent, so punitive damages should not be awarded in these cases (assuming the wikipedia text is correct and complete).
Hardware-support is a no-brainer. It's really simple: *do your research before you buy*, and it will be equally well supported in Linux.
Yeah, right. We live in a world where vendors change the chipsets in their cards without changing the model numbers. No amount of research will tell you whether or not a Belkin F5D7200 will work under Linux: it could be one of two entirely different cards, one of which works fine, the other of which doesn't, and you won't be able to tell until you get it out of the box. And if you're not very careful with the research you do, you'll just get the "it works fine" response, because it's listed on the hardware compatibility lists.
And when that's the only card that your local shops stock that does the job, are you supposed to travel further to a different shop? Where do you draw the line at what's convenient and what isn't?
The point is, if you run Windows, you can just buy hardware, take it out of the box, plug it in, stick the CD in the drive and expect it to work. If you run Linux, about 60% of the time it's even easier. But the other 40% is a real hassle.
Can you tell me how price fixing and other anti-competitive practices did this?
Easy. By consolidating the market and forcing it into a virtual monopoly, they've been able to substantially cut their prices in real terms (the cost of DOS + Windows at the time of Windows' release was about $150, which is roughly the same as a copy of WinXP Home, so there's 20 years worth of inflation that has come off that figure). Low prices make the PC platform look more attractive than alternatives (like Macs).
While obsoleting your computer every two or three years, they also make sure no one can sell you a new one that does not put money in their pocket.
That's funny. I'm sitting here writing this using a computer that's over 6 years old, yet I'm running the latest version of Windows without much difficulty. Also, my local computer store is quite capable of selling me a computer without any MS software, thanks.
In the early days, MS helped compatible system builders substantially by building customized versions of DOS (and even early Windows versions) that would run on machines that weren't 100% BIOS/hardware compatible with IBM. By doing this, they stimulated the competition in a way that wouldn't have been possible if, say, IBM had produced their own OS for the PC.
Although I still wonder where we'd be right now if IBM had bought CP/M 86 instead.
I hope you do realize that there's a difference between "spyware", "virus" and "worm". Hint: "spyware" is usually installed with the user's unknowing "consent".
Actually, I'm wondering whether you do. "Spyware" (and its annoying cousin, "adware," which randomly pops up adverts while you're trying to work) relates to what the software does when it is on your computer, whereas virus, worm, and trojan horse are all methods of distribution, and could perform any function desired once they have reached the computer (although by the definition of virus and worm the function must at least include spreading the malware further).
The spyware you're talking about is distributed in trojan horse form, but there's no reason it can't be distributed as a virus or worm. It is also commonly distributed in a fashion which doesn't quite merit being called a worm (because it isn't capable of spreading itself after it has installed) but isn't anything else I know a name for: it sits on a web site (with the web site owners cooperation) and infects passing browsers that are vulnerable to particular attacks. This is how better security could help prevent spyware, by eliminating this installation vector.
Then there was Windows 2000, and it was more stable and still easy to use.
Windows XP could hav been a Windows 2000 service pack. A better themable UI, a minor IE update, some utilities to do things like registry snapshots that were useful, but always available as cheap third party tools. No big deal.
Well, Win2K = NT 5.0 and WinXP = NT 5.1, released only a year and a half later, so what were you expecting?
That said, a lot of the useful features that were supposed to be in Win2K from the start (particularly remote desktop and fast user switching) did get shoved back to WinXP, so I wouldn't want to have stuck with 2K.
XP SP 2 was the same, except the firewall was so bad you still needed a third party firewall. And yeah, spyware got more popular in the last few years, so you need antispyware tools now too.
OK, this I don't get. What is wrong with the XP SP2 firewall? It does the job its supposed to do: stop incoming connections to services that you haven't specifically authorised to accept connections from outside. And please don't tell me it's because it doesn't do egress filtering: egress filtering is just about useless. It's trivially easy for any malware that wants to send data outbound to do so using a method that could not possibly be distinguished from legitimate traffic.
And yeah, spyware got more popular in the last few years, so you need antispyware tools now too.
Funnily enough, I've never had a problem with it. And I've never needed a tool to get rid of it off other people's systems, either.
As long as you don't go around installing random software from unknown sources, you won't have a problem. Of course, Linux users don't do this, because unknown sources don't tend to have a Linux version of their software available, so it isn't really an issue there, either.
Sure, you could install all this stuff in Windows, but you have to find it and pay for it and reboot and reboot and reboot.
Everything you list is trivially easy to find, completely free and doesn't require reboots to install under Windows these days.
There's even useful shit like strace in the OS.
Similar tools are available for windows as parts of the various SDKs, or for independent download. The rationale for not including it is that it's useless to anyone who isn't a developer, which seems sensible to me. Linux installs are often intimidating to new users, because they get to the software selection screen and see all the packages that are included, and have no idea how to choose what they need and what they don't need. Windows works the other way around: give the user a basic install with some tools that a large proportion of users will find useful (but without asking what they want), and then let them get anything else afterwards. It's a difference in philosophy and there advantages each way. But what it isn't, really, is a big deal.
Scenario A: I download an open source software, version 0.7.8.1 that does 80% of what I need. The project is either not in active development 'cause the developer got bored, or he/she/they refuse to add the features I want. I then look for a freelance programmer. I can expect at least 3 months of work; even at $8/hour, that's still $3840 for someone who works 40 hours per week.
Your estimate of the amount of work involved for fixing the problems you have with it is over the top. Chances are that your v0.7.8.1 program hasn't had the 480 hours of time that you reckon you'll need put into it at all, and that's probably between 2 or 3 programmers -- your single programmer with a fixed goal will be able to work more efficiently, because he doesn't need to communicate with other programmers to find out how they think things should work, integrate with modules they've written while he's been working on something else, etc: he just has to work to your spec.
Case in point: for years we've been using at work a Linux Debian as firewall / gateway / DHCP server. Every time we needed a new feature (VPN, bandwidth management) we needed to call a specialist. The box itself was free, but each configuration cost us about $250 [...] after paying $2500 I went out and bought a Zyxell Zywall 10W with wireless for $200, that does 98% of what I need.
You see, the problem with what you did here is that you tried to use a non-commercial Linux distribution in a commercial environment. While I still wouldn't recommend it if the dedicated device does what you need, a better way of doing this would have been to get a commercially supported Linux distribution (e.g. SuSE Server - $350); this would come with installation support to walk you through setting up what you need. If you later need additional support (e.g. because you need to add new features) you can purchase it at $900 per annum. This is a much cheaper way of doing what you wanted.
Since then I keep asking myself: why couldn't be Linux so easy to use, with clearly organized menus for everthing?
Many versions of Linux do come with such menus for a large portion of their features. Again, you made a mistake choosing Debian: its target market is geeks.
There's no convincing them to drive on the right though.
:)
But... if we were to drive on the right, and a highwayman were to attack us while he passed us coming the opposite way, we would have to defend ourselves with our left hands, and that's just not cricket!
Isn't that already the case ? If the domain isn't specified, the DNS server will check if there's a machine of that name in the current domain.
Yes, but most computers on the Internet these days don't actually know what their local DNS name is (if they have one at all, that is). That system stopped being generally useful sometime around '96 or '97.
I think you're right.
Sequential logic is hard to understand, and as project complexity gets higher it becomes exponentially more difficult to comprehend the possible interactions and determine what will happen in any particular circumstance.
The potential for research in modular non-sequential programming systems is great: a reliable, easy-to-use toolkit along these lines would become very popular very quickly, as it would make programming a lot easier than it is right now.
I suspect it is a long way in the future, though.
Sorry, region coding and control disabling (which is what prevents you from skipping the adverts) are not the same thing at all. I think HDDVD will still have the latter. They like that feature.
True, but the point is, because you've standardised on a single supplier, you automatically know who to call when you have a dead machine.
If you've been switching suppliers frequently to get the best deal on any particular component, you'll have to look up in your records to find out who you got that particular board from. And heaven help you if two different machines have had their labels confused at some point...
Fair enough, but I think google buys a lot more computers than the OP. Also, the PCs they need are very specialised, and they probably can't get a good deal on them from a mainline supplier like Dell, who typically package components google wouldn't need with all machines (CDROM, floppy drive, keyboard, mouse).
Its not only the price for the systems that randomly changes. I was on a client site recently while the client ordered some new hardware; about 15 Dell PCs. The quote came back roughly what he was expecting for the hardware (this was the second batch of similar machines he had ordered), but the delivery charge was astronomical. He called his sales rep, who informed him that they now charge a per-PC delivery charge, not a flat one. He cancelled the order because he would have to get clearance from his boss... the next day a new quote turns up with the same delivery charge as the original batch.
I've looked at those in a little detail, and while I'm not an expert I think I'm correct in the following summary:
1. The first is a patent on the long filename support of Win95 and later versions. It covers the method used for storing the long and short filenames in the same directory and for hiding the long filenames from older versions of DOS. , along with a few simple variations.
2. The second is a broader patent which just covers the same ideas but breaks them down into a larger number of simpler claims (the second being rather amusing: "2. The method as recited in claim 1 wherein the long filename contains more characters than the short filename." Well, duh!)
3. The third is for an API that allows DOS programs to use long filenames. Many of the claims are ridiculously broad and should easily be defeated by prior art claims:
Claim 1 basically reads "any extension to MS-DOS for using 64 bit file times". This is patenting the problem, not the solution, and therefore shouldn't stand.
Claim 2 is "the above, but using the same interface numbers we've used". The interface numbers chosen are an inconsequential detail, and do not satisfy any reasonable definition of non-obviousness.
Claim 3 is "any operating system that implements both short and long files and can convert 64 bit file times to and from BCD." To sink this one, you need to find an OS that predates Win95 that allowed you to create both short and long filenames referring to the same file. I'd be surprised if this hadn't been done before MS did it. The date conversion features don't seem to satisfy the criterion of non-obviousness to me.
Claim 4 is "any operating system that automatically creates a short name when you tell it to create a file with a long name." This might be the hardest claim to sink. They might actually have invented this feature.
Claim 5 is the same, except you use "int 21h" as the syscall function (i.e. the OS is DOS or CP/M 86 compatible)
Claim 6 is the same as 5, except the function number is 7139h.
Claim 7 is the same as 4, except it's for deleting files, rather than creating them.
Claims 8 and 9 are equivalent to 5 and 6.
Claims 10-12 are a minor variation on claims 7-9.
Claims 13-15 are the equivalents for renaming files
Claims 16-19 are the equivalents for listing directories
Claims 20-24 are the same as the above claims 4, 7, 10, 13 and 16, except that they are for disks with the software stored on them, rather than a computer with the software loaded.
It's worth noting that all of these were originally filed prior to the release of win95, but the application was abandoned and a new application filed at a later date.
Hell, that's hardly the most egregious error in the document. Apparently, the FAT format is used by Samba, the open source software that allows Linux to share files with Windows and read files from Windows disks.
Where'd I leave the cluestick?
I kind of agree with your argument. There are just two problems with it:
1. History has shown that people with strong leanings in any direction are very frequently unable to overcome those leanings when interpreting facts before them. Two different people can read the same facts in entirely different ways because of this.
2. If a case were as simple as reading the law and systematically applying to it, it would not reach the supreme court. Cases that get as far as the supreme court are those that require interpretation of the more complex aspects of law, especially the interactions between different laws that may have been written with different goals in mind. This interpretation can frequently be a highly personal matter. This is why many supreme court judgements are not unanimous.
Because of these facts, I think it is best if all judges have no strong leanings towards any particular political viewpoint, other than those that are actually embodied in the laws of the land.
IIRC, there wasn't really a bug in the disk compression; the issue was lazy writes. Users who, for example, shut the system down precipitously, might have had written data lost, but the same is true of any modern OS. I think all 6.2 did in this regard was to shut off lazy writes by default.
Right. The problem is, with DOS, how do you know when the data has been written to disk? Turns out the common recommendation was to wait until you got the DOS prompt back after quitting your apps, but this wasn't adequate protection. The fix was to make command.com flush the disk before displaying the prompt every time.