I am not sure why you say that mathematics is exceptional in that its point is publication. The same is true of all scientific disciplines I can think of right now.
But you weren't comparing mathematics to other scientific disciplines, you were comparing it to things like investment banking and applied statistics and engineering. You were talking about areas where publication of the results is not an essential part of the process.
It really doesn't matter if you accomplish the verification of these steps in your favorite package or "roll" your own objects in python, ruby, perl, etc.
Just as in other academic disciplines, you need to document what you did in a way that other people can verify the results. If an essential part of the operation is running a program, then you need to document that part of the operation in a way that allows people to verify it. If that step depends on software that isn't generally available or can't be verified, then you haven't documented it.
This is reproducible because the steps are documented and anyone can write their own program to check these property for every element of the set.
The article in the AMS that this is all about was specifically raising the question of whether this will remain true. You can't challenge that question simply by stating its opposite.:)
I guess that depends on how conversational a style you have in email. I mean, both are logged and on OSX at least Spotlight brings both email messages and IMs right up. I don't find any kind of text communication really similar to conversation... not as long as the backspace key works.
In any case, I'm not saying that they're "the same", I'm saying that the idea that IMs are more immediate and attention-getting than email depends on the perspective of the recipient. It's not inherent in the medium.
In fact it might be interesting to have a gateway between the two, so IMs get turned into email and replies go back out as IMs, and see how that changes things.
I am definitely finding it more difficult to work closely with people who just have e-mail.
But is it because of the medium or the person? Some people just don't want to operate in "real time", and IM clients tend not to support a "silent away" mode where messages are just queued up. If they found a client that let them respond asynchronously they'd use that... but you'd have just as much trouble getting hold of them as you do through email.
On the other hand, email clients can be made to bring up an alert within seconds of you sending a message... so you can effectively use both in "real time" mode.
This is more likely explained by a variant of Hanlon's Razor, to wit "never attribute to malice what could be explained by laziness".
Since they know there's only one instance of the browser running on the phone, this is an easier way to maintain session information than using cookies. It's cheap, cheesy, and lazy.
On to the solution: it shouldn't be hard to create a Mach/Cocoa overrider (using any of the various tricks to patch running apps, like APE does) to change the IMEI seen by widgets if you really want to hide which phone you are.
I mean, really, if you've only been on XP a couple of years, what's the benefit of upgrading to Vista?
XP was released in 2001
Windows 2000 SP4 was released in 2003
Windows 2000 "SP5" ("rollup 1" for SP4) was released in 2005, because many companies were still using 2000 in 2005.
Windows 2000 extended support continues through 2010
Vista was released in 2006
Windows XP is going to have at least one more service pack in 2008
I predict that Windows XP will have a service pack or a rollup in 2010, and companies running XP now don't really need to face any kind of forced upgrade to Vista or Windows 7 until then.
Windows XP extended support will continue beyond that point.
If they're using email effectively it won't matter whether you send them an email or an IM, they'll get it just as quickly. It's got nothing to do with IM being better for getting people's attention, it's got to do with them simply not wanting to use email.
I've worked with people who insist on IM, instead of email, and they generally ignore their email, often for days. So I use whatever they want to use, but if I'm busy I'll turn the IM and my email program BOTH off. I'm just as accessible either way, and just as inaccessible, as well.
But the same is true for most of academic research.
Except for one thing. The point of mathematical research is publication. If you're publishing into that corpus, whether you're doing it because that's what you do or because you ran across something as a side effect of something else (and that's the rare exception, the exceptions you describe are famous BECAUSE they are exceptional), if you're actually publishing you're operating in the pure math world.
And the point is that IF you are publishing a paper, you are *by definition* operating under what you're calling "open math"s rules, and one of those rules is that you document everything needed to allow someone else to reproduce your results. If they're not reproducible, they're not results. It doesn't matter how you discovered the results, whether it's serendipity or persistence, if one of the steps is "use this magic software", then you shouldn't be publishing them as pure math.
Pythagoreans were very secretive. So were statisticians in the 19th century. I am pretty sure investment bankers do a great deal of math that they don't want anyone to ever see because it gives them an edge in the market.
I'm pretty sure investment bankers aren't publishing in mathematical journals.
This isn't about applied math, this is about theory. Investment bankers don't need to produce proofs, let alone publish them... the results are all that matters. Investment bankers and statisticians aren't pushing the boundaries of mathematics, the way you imply, they're applying the framework created by pure mathematics.
This isn't like gunpowder, at all. The practical application of most pure math is decades after its discovery, if at all. Even the closest discipline to "practical pure mathematics", cryptography, is more a matter of mining theorems looking for operations that have properties useful to crypto... and the underlying work is based on applications of well known work.
But for all that, crypto has had to accept that it's most effective and most useful when it's published and open. That's just what has been found to work best for pure mathematics... which is why it's considered "natural".
You're saying that contextual menus aren't evidence of non-single-button thinking because they're available via a single button as well.
I'm saying it because they were introduced over three years before Apple supported multi-button mice at all, and two years before USB support opened up the possibility of third-party multi-button mice that coule work with something like USB Overdrive to get right-click contextual menus. I was actually looking for ADB two button mice for OS 8 and the only ones I found had the right button hard-coded in the mouse itself and couldn't be made to support contextual menus. The one I still have makes it a "click lock" for dragging.
This is what you call "cause and effect". A user interface feature is not designed for input devices that weren't available until years after the feature was introduced.
On the other hand, all PC mice with very few exceptions (none that I can recall) already had at least two buttons before Windows 95, and contextual menus in Windows 95 didn't even support a modifier key. To get them without right-clicking you needed to select the object first then use shift-F10 or (if you had a "Windows keyboard") the "menu" key.
This is, again, "cause and effect". A user interface feature is designed for input devices available when it was introduced.
Therefor should a worm ever make its way onto one of the older versions, you'd be wanting to upgrade right quick.
I'm still using Windows 2000, and have yet to suffer from malware on it despite eschewing antivirus CPU hogs and only updating when I actually need updated drivers or other new functionality available in updates.
So why do you suppose I should be concerned about malware in OS X?
The biggest reason that Windows is so attractive to malware is because of the behavior of one component of the system: the HTML control. If you are careful with firewalls and avoid using any application that uses the HTML control on untrusted content, then the vast majority of remote exploits will continue to pass you by. Now I don't recommend actually making do without antivirus for most Windows users - I've got the experience of 20 years as a network administrator to guide my computer use - but if you avoid using Microsoft's malware incubators you're pretty solid.
Anyway, yes, there are reasons to update. But it's not compulsory, it's not even always desirable (I held off on Panther because the new Finder just put me off my feed, and Windows XP still seems like a downgrade from Windows 2000), and there's no reason to obsessively update to every new version, and totting up all the releases of OS X as if that WAS necessary is really really misleading.
I dislike that kind of false accounting no matter who does it, whether it's Windows nuts overstating the cost of OS X, or Apple fans trying to make Macs look as cheap as Wintel boxes.
Right-click menus are ubiquitous throughout the OS
Those are "context menus", not "right-click menus", and may be activated by control-click, click-and-hold, and right-click... the first two operations are designed around a single mouse button, and the third was only made standard this century.
This design means that contextual menus can not be activated on a drag because click-and-hold is incompatible with dragging, and so it would not be consistent to implement that operation.
Which is why I said this flaw was mitigated by the single-mouse design... it's harder to accidentally trigger it than on Windows.
Good for you. I tried that, and ended up buying a program that lets me designate corners for right-click. It's still far from perfect, but it's light-years better than two-finger clicking.
No, it's one button with a capacitance sensor to detect where on the button you're clicking. This test is unreliable for a number of people, and defaults to left-click. If you lucked out, or have been able to train yourself to use it successfully, congratulations.
Shift, control, command, and option are on the keyboard, not the mouse.
They are, however, essential to the operation of the mouse. You can't operate the mouse on a Mac without using the keyboard, so the modifier buttons on the keyboard are (and have been, since the '80s) stand-ins for extra mouse buttons.
Working at the same time (chording) is not working independently, that is working in unison.
"working independently" means "the state of one does not have an effect on the state of the other".
I'm not sure it's fair to say that the current OS X interface is designed from a single-button mouse perspective.
It really is. The right click is just one of three alternate mechanisms to bring up a context menu: click-and-hold, control-click, and right click. The support for this is patchy: custom widgets have to explicitly handle all three cases, and the right click won't activate a window.
Windows right-drag does follow the same philosophy. You right-drag and THEN get a menu to select how you want to do the move, and I've seen people who right-drag all the time. This makes it easier to accidentally trigger the bug.
Windows 3.1 did nothing with the second button because, at the time, nobody knew what to do with it.
Actually, that's not true. The original window GUI invented by Xerox used the right button for "menu", and by this time Sun and other workstation manufacturers were also using the right button for this purpose, as was Amiga. The middle button was used for a variety of different purposes, but the left-for-select-right-for-menu design goes back to the late *70s*.
Apple rejected this design in favor of a single button, which meant that they effectively moved the extra buttons from the mouse to the keyboard... which is why I describe the current incarnation as a "five button mouse" with click, control-click, shift-click, command-click, and option-click.
A local firewall isn't normally necessary on a UNIX system, since there should be no required services that can't run without leaving a promiscuous listening TCP port, so a firewall isn't necessary to protect local services from remote exploits.
I have only ever used a local firewall on any UNIX system when I'm also performing packet forwarding (ie, acting as a router) and so can't control access at the application layer.
So the main purpose of a local firewall on UNIX is not to protect standard services from attack, it's to prevent a backdoor listener from being accessed. Which is what this does.
The real defect seems to be in the implementation not actually ensuring terminated services are shut down. That's a bug (though not a design defect).
I would argue that "deficient by design" would apply: it's missing useful functionality. But that doesn't make it defective.
The process you describe for moving a file in your post is so basic that it should be child's play to automate and combine it into a single function.
And yet the default behavior in Windows is the same as on the Mac. Funny thing, that.
The only difference is that on Windows you can drag with a different button to change the behavior, where on the Mac you have to hold down a meta-key (which also works on Windows, by the way). This is where Apple lucked out: it's harder to accidentally trigger the bad behavior than it would be if they'd adopted multi-button mice earlier.
News to you apparently but, plug in an n-button USB mouse and for the last decade or so, It Just Works.
Yes, I know, I use a Microsoft optical mouse on my Mac.
Now, plug in a 47 button USB mouse on your Mac. Having done that tell me how you drag files from one volume to another and thus trigger this bug using only the mouse? You can't do it. You have to deliberately hold down a meta-key on the keyboard while dragging to force OS X to MOVE rather then (as it does by default) COPY the files.
The mouse that comes with an iMac actually has four buttons.
No, it's got 7. The mouse button, the scroll ball, the squeeze button, and the shift, control, command, option chords.
That's beside the point, in any case. The issue is not whether Apple currently ships with a 1, 2, 4, or 8 button mouse, but that the user interface is designed for a single button mouse, which (in this case) is actually helpful because it avoids the possibility of a normal drag operation triggering this bug.
It looks like one button, but both the left and right sides click independently.
No they don't. You can't click both at the same time (chording), and if you click on the right side without making sure that no part of your hand is touching the left side near the front it registers as a left-click.
If you had a bought a Mac with 10.0, then the costs would be rather different if you upgraded each time.
Why would I have upgraded each time?
I'm still running Jaguar on one computer, and I'm still running Windows 2000 on 3 of them.
This is what the grandparent was talking about, not your misrepresentation of it.
I'm not "misrepresenting" anything. I'm challenging the myth that just because Apple, Microsoft, Red Hat, Palm, FreeBSD, or any other software publisher releases an update that you HAVE to update.
The upgrades I made were OS 8 -> 10.2, OS 9 -> 10.2, 10.1 -> 10.2, Windows 3.11 -> Windows 2000, Windows 95 -> WIndows 2000, and two Windows 98 -> Windows 2000. I also bought two Macs running Panther, a laptop running Tiger, and a laptop running Windows XP.
Where the hell does this fantasy that it's necessary to upgrade to the latest OS come from? Nobody thinks its' strange if you didn't upgrade from VHS to laserdisc to DVD but just skipped the laserdisc. Nobody thinks it's strange if you don't buy a new car as soon as the old one is paid off. Nobody thinks it's strange, even, if you buy a *used* car that's older than the leading edge.
I only upgrade my computers, cars, or anything else if there's something I want to do that requires me to upgrade.
All apple computers now ship with two button mice, and have for a while.
But the user interface is defined in terms of a single button mouse.
I cant see why it would be a flaw to have the default action of a drag and drop be a copy instead of a move.
The default action of a drag and drop in the situation where this flaw can occur *IS* a copy instead of a move. The only way to trigger the flaw is to hold down a meta-key while dragging.
It's only a move when it's on the same disk, and so the underlying operation really IS a move and not a copy-and-delete, and the problem doesn't show up.
I had to re-download and install Skype, and now I have to run it with the firewall switched off.
The firewall is not an essential component on a UNIX system the way it is on Windows, because you can actually turn off all listening ports and go "dead" without having to firewall off internal services that can't run without a TCP port open.
A computer system with no open ports is just as secure whether it's firewalled or not.
Luckily another design flaw in OS X makes it hard to trigger this bug. Because of the single-button mouse the only way to move files from one volume to another (rather than copying them) requires you to hold down some meta-key while dragging. If you just drag the files you get a copy.
Slashdot Mirror
I am not sure why you say that mathematics is exceptional in that its point is publication. The same is true of all scientific disciplines I can think of right now.
:)
But you weren't comparing mathematics to other scientific disciplines, you were comparing it to things like investment banking and applied statistics and engineering. You were talking about areas where publication of the results is not an essential part of the process.
It really doesn't matter if you accomplish the verification of these steps in your favorite package or "roll" your own objects in python, ruby, perl, etc.
Just as in other academic disciplines, you need to document what you did in a way that other people can verify the results. If an essential part of the operation is running a program, then you need to document that part of the operation in a way that allows people to verify it. If that step depends on software that isn't generally available or can't be verified, then you haven't documented it.
This is reproducible because the steps are documented and anyone can write their own program to check these property for every element of the set.
The article in the AMS that this is all about was specifically raising the question of whether this will remain true. You can't challenge that question simply by stating its opposite.
I guess that depends on how conversational a style you have in email. I mean, both are logged and on OSX at least Spotlight brings both email messages and IMs right up. I don't find any kind of text communication really similar to conversation... not as long as the backspace key works.
In any case, I'm not saying that they're "the same", I'm saying that the idea that IMs are more immediate and attention-getting than email depends on the perspective of the recipient. It's not inherent in the medium.
In fact it might be interesting to have a gateway between the two, so IMs get turned into email and replies go back out as IMs, and see how that changes things.
I am definitely finding it more difficult to work closely with people who just have e-mail.
But is it because of the medium or the person? Some people just don't want to operate in "real time", and IM clients tend not to support a "silent away" mode where messages are just queued up. If they found a client that let them respond asynchronously they'd use that... but you'd have just as much trouble getting hold of them as you do through email.
On the other hand, email clients can be made to bring up an alert within seconds of you sending a message... so you can effectively use both in "real time" mode.
This is more likely explained by a variant of Hanlon's Razor, to wit "never attribute to malice what could be explained by laziness".
Since they know there's only one instance of the browser running on the phone, this is an easier way to maintain session information than using cookies. It's cheap, cheesy, and lazy.
On to the solution: it shouldn't be hard to create a Mach/Cocoa overrider (using any of the various tricks to patch running apps, like APE does) to change the IMEI seen by widgets if you really want to hide which phone you are.
I mean, really, if you've only been on XP a couple of years, what's the benefit of upgrading to Vista?
XP was released in 2001
Windows 2000 SP4 was released in 2003
Windows 2000 "SP5" ("rollup 1" for SP4) was released in 2005, because many companies were still using 2000 in 2005.
Windows 2000 extended support continues through 2010
Vista was released in 2006
Windows XP is going to have at least one more service pack in 2008
I predict that Windows XP will have a service pack or a rollup in 2010, and companies running XP now don't really need to face any kind of forced upgrade to Vista or Windows 7 until then.
Windows XP extended support will continue beyond that point.
If they're using email effectively it won't matter whether you send them an email or an IM, they'll get it just as quickly. It's got nothing to do with IM being better for getting people's attention, it's got to do with them simply not wanting to use email.
I've worked with people who insist on IM, instead of email, and they generally ignore their email, often for days. So I use whatever they want to use, but if I'm busy I'll turn the IM and my email program BOTH off. I'm just as accessible either way, and just as inaccessible, as well.
But the same is true for most of academic research.
Except for one thing. The point of mathematical research is publication. If you're publishing into that corpus, whether you're doing it because that's what you do or because you ran across something as a side effect of something else (and that's the rare exception, the exceptions you describe are famous BECAUSE they are exceptional), if you're actually publishing you're operating in the pure math world.
And the point is that IF you are publishing a paper, you are *by definition* operating under what you're calling "open math"s rules, and one of those rules is that you document everything needed to allow someone else to reproduce your results. If they're not reproducible, they're not results. It doesn't matter how you discovered the results, whether it's serendipity or persistence, if one of the steps is "use this magic software", then you shouldn't be publishing them as pure math.
Pythagoreans were very secretive. So were statisticians in the 19th century. I am pretty sure investment bankers do a great deal of math that they don't want anyone to ever see because it gives them an edge in the market.
I'm pretty sure investment bankers aren't publishing in mathematical journals.
This isn't about applied math, this is about theory. Investment bankers don't need to produce proofs, let alone publish them... the results are all that matters. Investment bankers and statisticians aren't pushing the boundaries of mathematics, the way you imply, they're applying the framework created by pure mathematics.
This isn't like gunpowder, at all. The practical application of most pure math is decades after its discovery, if at all. Even the closest discipline to "practical pure mathematics", cryptography, is more a matter of mining theorems looking for operations that have properties useful to crypto... and the underlying work is based on applications of well known work.
But for all that, crypto has had to accept that it's most effective and most useful when it's published and open. That's just what has been found to work best for pure mathematics... which is why it's considered "natural".
You're saying that contextual menus aren't evidence of non-single-button thinking because they're available via a single button as well.
I'm saying it because they were introduced over three years before Apple supported multi-button mice at all, and two years before USB support opened up the possibility of third-party multi-button mice that coule work with something like USB Overdrive to get right-click contextual menus. I was actually looking for ADB two button mice for OS 8 and the only ones I found had the right button hard-coded in the mouse itself and couldn't be made to support contextual menus. The one I still have makes it a "click lock" for dragging.
This is what you call "cause and effect". A user interface feature is not designed for input devices that weren't available until years after the feature was introduced.
On the other hand, all PC mice with very few exceptions (none that I can recall) already had at least two buttons before Windows 95, and contextual menus in Windows 95 didn't even support a modifier key. To get them without right-clicking you needed to select the object first then use shift-F10 or (if you had a "Windows keyboard") the "menu" key.
This is, again, "cause and effect". A user interface feature is designed for input devices available when it was introduced.
Therefor should a worm ever make its way onto one of the older versions, you'd be wanting to upgrade right quick.
I'm still using Windows 2000, and have yet to suffer from malware on it despite eschewing antivirus CPU hogs and only updating when I actually need updated drivers or other new functionality available in updates.
So why do you suppose I should be concerned about malware in OS X?
The biggest reason that Windows is so attractive to malware is because of the behavior of one component of the system: the HTML control. If you are careful with firewalls and avoid using any application that uses the HTML control on untrusted content, then the vast majority of remote exploits will continue to pass you by. Now I don't recommend actually making do without antivirus for most Windows users - I've got the experience of 20 years as a network administrator to guide my computer use - but if you avoid using Microsoft's malware incubators you're pretty solid.
Anyway, yes, there are reasons to update. But it's not compulsory, it's not even always desirable (I held off on Panther because the new Finder just put me off my feed, and Windows XP still seems like a downgrade from Windows 2000), and there's no reason to obsessively update to every new version, and totting up all the releases of OS X as if that WAS necessary is really really misleading.
I dislike that kind of false accounting no matter who does it, whether it's Windows nuts overstating the cost of OS X, or Apple fans trying to make Macs look as cheap as Wintel boxes.
Right-click menus are ubiquitous throughout the OS
Those are "context menus", not "right-click menus", and may be activated by control-click, click-and-hold, and right-click... the first two operations are designed around a single mouse button, and the third was only made standard this century.
This design means that contextual menus can not be activated on a drag because click-and-hold is incompatible with dragging, and so it would not be consistent to implement that operation.
Which is why I said this flaw was mitigated by the single-mouse design... it's harder to accidentally trigger it than on Windows.
Good for you. I tried that, and ended up buying a program that lets me designate corners for right-click. It's still far from perfect, but it's light-years better than two-finger clicking.
The mouse buttons is two buttons, not one.
No, it's one button with a capacitance sensor to detect where on the button you're clicking. This test is unreliable for a number of people, and defaults to left-click. If you lucked out, or have been able to train yourself to use it successfully, congratulations.
Shift, control, command, and option are on the keyboard, not the mouse.
They are, however, essential to the operation of the mouse. You can't operate the mouse on a Mac without using the keyboard, so the modifier buttons on the keyboard are (and have been, since the '80s) stand-ins for extra mouse buttons.
Working at the same time (chording) is not working independently, that is working in unison.
"working independently" means "the state of one does not have an effect on the state of the other".
I'm not sure it's fair to say that the current OS X interface is designed from a single-button mouse perspective.
It really is. The right click is just one of three alternate mechanisms to bring up a context menu: click-and-hold, control-click, and right click. The support for this is patchy: custom widgets have to explicitly handle all three cases, and the right click won't activate a window.
Windows right-drag does follow the same philosophy. You right-drag and THEN get a menu to select how you want to do the move, and I've seen people who right-drag all the time. This makes it easier to accidentally trigger the bug.
Windows 3.1 did nothing with the second button because, at the time, nobody knew what to do with it.
Actually, that's not true. The original window GUI invented by Xerox used the right button for "menu", and by this time Sun and other workstation manufacturers were also using the right button for this purpose, as was Amiga. The middle button was used for a variety of different purposes, but the left-for-select-right-for-menu design goes back to the late *70s*.
Apple rejected this design in favor of a single button, which meant that they effectively moved the extra buttons from the mouse to the keyboard... which is why I describe the current incarnation as a "five button mouse" with click, control-click, shift-click, command-click, and option-click.
You can map your right mouse button on a Mac to be the same meta-key that would MOVE instead of COPY.
Not without third-party software. Again, my point was that this is not something that can happen by accident on the Mac, the way it can on Windows.
But then again, why would you want to make a Mac more like a PC?
Since my point is that the fact that they're not alike is an advantage, why do you think I want to?
A local firewall isn't normally necessary on a UNIX system, since there should be no required services that can't run without leaving a promiscuous listening TCP port, so a firewall isn't necessary to protect local services from remote exploits.
I have only ever used a local firewall on any UNIX system when I'm also performing packet forwarding (ie, acting as a router) and so can't control access at the application layer.
So the main purpose of a local firewall on UNIX is not to protect standard services from attack, it's to prevent a backdoor listener from being accessed. Which is what this does.
The real defect seems to be in the implementation not actually ensuring terminated services are shut down. That's a bug (though not a design defect).
I would argue that "deficient by design" would apply: it's missing useful functionality. But that doesn't make it defective.
The process you describe for moving a file in your post is so basic that it should be child's play to automate and combine it into a single function.
And yet the default behavior in Windows is the same as on the Mac. Funny thing, that.
The only difference is that on Windows you can drag with a different button to change the behavior, where on the Mac you have to hold down a meta-key (which also works on Windows, by the way). This is where Apple lucked out: it's harder to accidentally trigger the bad behavior than it would be if they'd adopted multi-button mice earlier.
Now you're making it sound like MacOS is copying Windows 3.1.
The multi-button mouse comes from Xerox: Smalltalk, Interlisp-D, and the Xerox Star office system.
News to you apparently but, plug in an n-button USB mouse and for the last decade or so, It Just Works.
Yes, I know, I use a Microsoft optical mouse on my Mac.
Now, plug in a 47 button USB mouse on your Mac. Having done that tell me how you drag files from one volume to another and thus trigger this bug using only the mouse? You can't do it. You have to deliberately hold down a meta-key on the keyboard while dragging to force OS X to MOVE rather then (as it does by default) COPY the files.
The mouse that comes with an iMac actually has four buttons.
No, it's got 7. The mouse button, the scroll ball, the squeeze button, and the shift, control, command, option chords.
That's beside the point, in any case. The issue is not whether Apple currently ships with a 1, 2, 4, or 8 button mouse, but that the user interface is designed for a single button mouse, which (in this case) is actually helpful because it avoids the possibility of a normal drag operation triggering this bug.
It looks like one button, but both the left and right sides click independently.
No they don't. You can't click both at the same time (chording), and if you click on the right side without making sure that no part of your hand is touching the left side near the front it registers as a left-click.
OK, how do you drag files from one volume to another, triggering this bug, without holding down a meta-key?
(and how is pointing out that it's a minor problem FUD?)
If you had a bought a Mac with 10.0, then the costs would be rather different if you upgraded each time.
Why would I have upgraded each time?
I'm still running Jaguar on one computer, and I'm still running Windows 2000 on 3 of them.
This is what the grandparent was talking about, not your misrepresentation of it.
I'm not "misrepresenting" anything. I'm challenging the myth that just because Apple, Microsoft, Red Hat, Palm, FreeBSD, or any other software publisher releases an update that you HAVE to update.
The upgrades I made were OS 8 -> 10.2, OS 9 -> 10.2, 10.1 -> 10.2, Windows 3.11 -> Windows 2000, Windows 95 -> WIndows 2000, and two Windows 98 -> Windows 2000. I also bought two Macs running Panther, a laptop running Tiger, and a laptop running Windows XP.
Where the hell does this fantasy that it's necessary to upgrade to the latest OS come from? Nobody thinks its' strange if you didn't upgrade from VHS to laserdisc to DVD but just skipped the laserdisc. Nobody thinks it's strange if you don't buy a new car as soon as the old one is paid off. Nobody thinks it's strange, even, if you buy a *used* car that's older than the leading edge.
I only upgrade my computers, cars, or anything else if there's something I want to do that requires me to upgrade.
Why is that so bloody strange?
All apple computers now ship with two button mice, and have for a while.
But the user interface is defined in terms of a single button mouse.
I cant see why it would be a flaw to have the default action of a drag and drop be a copy instead of a move.
The default action of a drag and drop in the situation where this flaw can occur *IS* a copy instead of a move. The only way to trigger the flaw is to hold down a meta-key while dragging.
It's only a move when it's on the same disk, and so the underlying operation really IS a move and not a copy-and-delete, and the problem doesn't show up.
I had to re-download and install Skype, and now I have to run it with the firewall switched off.
The firewall is not an essential component on a UNIX system the way it is on Windows, because you can actually turn off all listening ports and go "dead" without having to firewall off internal services that can't run without a TCP port open.
A computer system with no open ports is just as secure whether it's firewalled or not.
Luckily another design flaw in OS X makes it hard to trigger this bug. Because of the single-button mouse the only way to move files from one volume to another (rather than copying them) requires you to hold down some meta-key while dragging. If you just drag the files you get a copy.