You're going to write it down after, but the doc's talking fast enough, you don't have the writing skill to do it on paper.
Not to detract from the rest of your argument, which I agree with, but most people can't type faster than (or even close to as fast as) they can write - they'll write something down rather than type it into notepad.
So a Powerbook G5 that runs at 1.6GHz won't run normal apps much faster than a PBG4 at the same clock.
I think you'll find the ~5x improvement in bus bandwidth will give a quite healthy boost in real-life performance, even if the clock speed remains the same.
Re:Looks like they've got their focus...
on
Longhorn Preview
·
· Score: 1
That was my impression too, yet the "administrator by default" installation setting hasn't changed, even on the latest batch of WinXP SP2 CDs.
Minor changes/in Windows/. Applications may need major reworking - but that's not Microsoft's fault and there's nothing they can do about it.
What I'm trying to say here, is that everyone goes on about how Windows needs to be "redesigned" or "rewritten from the ground up" to be "secure" when all it really needs are a few default settings tweaked. The problem for Microsoft is that tweaking these defaults is potentially going to piss off a lot of people.
In fact, the only time I log on to her machine as admin is to use Windows Update or to install software.
If you're not already, you should really use "Run As" for this sort of thing.
The OSI model has nothing whatsoever to say about how Windows printer adds work. Trust me on this one. Why do you think that the OSI model implies that an IP printer should be considered "local" as opposed to "network"?
Stop moving those goal posts around. Your original claim was that Windows was not "designed with IP in mind" based on the rather shaky assertion that all IP-related configuration was "complex". This has since devolved to basing your far-reaching criticism solely on the UI for setting up IP printers (a relatively uncommon task, even in networked environments).
I hope you realise how silly this line of reasoning is. It's like saying "Linux wasn't built with SCSI in mind" because fdisk is hard to use.
In any event, my comment on the OSI model was because you are trying to compare something at the network layer (IP) with something at the session layer (NetBIOS). NetBIOS runs *on top of IP* (or NetBEUI, or IPX, or probably anything else you want it to).
Assuming you _really_ mean the old, unroutable NetBEUI protocol (that operates at the same level as IP), and not NetBIOS, it hasn't since Windows 98 (and even then, I seem to recall Windows 98 defaults to NetBIOS over IP - but it's been a long time since I've installed Windows 98).
NT variants of Windows have preferred NetBIOS over IP (as opposed to NetBIOS over NetBEUI, which is what I think you're talking about) since at _least_ NT4 (probably even earlier, but it's been a very long time since I've installed earlier versions of NT).
However, all this is irrelevant - Windows is no less (or more) "designed to use IP" than OS X is "designed to use Firewire". IP is simply a minor implementation detail of getting network data from machine A to machine B, just like Firewire is a minor implementation detail of how to connect a hard disk to a computer.
The ad-hoc filesharing which was mentioned earlier in the thread (right click, share this folder): what is that based on?
SMB (or CIFS, depending on what you want to call it). Again, complete independent of the network layer - it can run over IP, NetBEUI, IPX/SPX, etc (just like AFP on MacOS).
However, consider the paradigm: in Unix, EVERYTHING is manually editing obscure text files.
Ah, but some are more obscure than others. By your reasoning, some Linux distributions aren't "designed with IP in mind" and others are, despite being the same OS, merely because they happen to have a more complex UI for IP configuration.
Would you say that windows 95 was "built with IP in mind," or was IP "grafted on"?
Ditto for NT.
I would say it's completely and utterly irrelevant. Both of them were built to be network OSes. It's like asserting an OS was "built with SCSI in mind" or "built with AGP in mind" or "built with SDRAM in mind" whereas others had it "grafted on".
Again, I repeat my claim that IP on windows is in the Mac OS 8 realm: it works, but it's clunky, and it was treated as an afterthought when the OS was written.
And I repeat that basing your criticisms of low-level OS design on one example of clunky *user interface* is idiotic.
Why didn't anyone mention to the DOJ that Mickeysoft should release 'driver source code'.
It's worth noting that the only platform that "requires driver source code" is Linux - everyone else's is stable enough such that hardware vendors can write their own drivers and not have to worry about the next sub-minor kernel revision breaking them.
Some (most, IMHO) programming is just another trade - like carpentry, plumbing, etc. This is the stuff where the same thing just gets done over and over and over again by different people in different places but all using the same basic principles/knowledge and tools (algorithms, languages, IDEs, etc).
Some programming would qualify as engineering - carefully specified, carefully implemented, deeply tested, task-specific.
Some programming would qualify as hobby - the people who try to roll together [some program] with [some wierd constraint] just for the hell of it.
I can't say I'd consider any sort of programming "art".
Re:Looks like they've got their focus...
on
Longhorn Preview
·
· Score: 1
That's not true at all. My default WinXP installation CD prompts me for a user account, and by default, makes it an administrator. Without subtly forcing the users to create a non-admin account, there's nothing preventing Joe Average from running as an administrator all of the time.
But - as I said - there's nothing stopping them from running as a regular user all the time either, except poorly written apps.
In this case, it's the operating system.
It's a minor default configuration detail. Hardly anything that requires major changes.
Re:The greatest fear of the Open Source community
on
Longhorn Preview
·
· Score: 1
Even if that were to happen, i.e. MS basically giving up, Linux is still a Unix system, and Windows is Windows.
This is irrelevant to anyone not already indoctrinated into unix.
That means that i can type "2005 yesemite vacation" 89 times in 89 "get info" windows... versus that clumsy old way of putting 89 files into a folder called "2005 yosemite vacation".
For some reason I'd be willing to bet they'll have an "update all selected files with [metadata]" in a file properties box.
We just have trouble making ourselves pay $200+ for an OS that is just as insecure as the one we already are using daily only to get extra "eye-candy".
Windows - like Linux and OS X - is only as insecure as you choose to make it.
I'm always surprised that this isn't touted more with Linux / Mac, etc. The fact that you have to reboot to do most upgrades is a pain.
That's because out in the real world, very few people care about rebooting (most of them shut the machine down every night anyway).
Re:Headlines running together in my head
on
Longhorn Preview
·
· Score: 1
On OS X I have to type in my admin password to fubar anything.
Actually, if you're an "Admin" user, you have write access (without any password prompt) to quite a bit of the system - probably everything in/Applications, for example (unless you've gone through an manually twiddled file permissions).
A virus could propogate quite nicely through an OS X system - assuming it could get itself executed by an "Admin" user (very high possibility) - simply by infecting nearly every application installed on the machine. For example (and I'm speaking hypothetically without fully researching the feasibility) I suspect it would be possible to have an Admin-run virus infect Installer.app and then use that as an avenue to achieve root privileges the next time some app's install procedure pops up the password box.
Re:Mystery of the computer industry
on
Longhorn Preview
·
· Score: 2, Insightful
Is it really that high-tech firms are full of dorks without any taste ? Is the difference with Apple the fact that Steve Jobs decides, and the guy actually has good taste ?
No, it's because different people have different tastes.
Does this mean we will get a translucent "Screen of Death" superimposed over the crashed OS screen, instead of the solid-blue one?
Depends on whether Apple has a patent on their "translucent kernel dump" screen.
Re:Looks like they've got their focus...
on
Longhorn Preview
·
· Score: 1
That is extremely debatable. While inexperienced users can be tricked quite easily, you'd think that the biggest corporation on the planet would at least be able to limit the problem.
How ? They don't write every application. They can't _make_ users not do stupid things.
The firewall in SP2 was a start, but how about making something that stops the OS being taken over after one wrong click on a message box in IE?
You can do that now - just don't run IE as an Administrator.
There are so many people with spyware/virus problems, including reasonably experienced users, that at some point Microsoft has to start taking the blame.
There are a handful of viruses that actually exploit OS vulnerabilities and bugs (most - if not all - of which have been fixed) everything else comes back to either the user (deliberately running malicious code, running with higher than necessary privileges) and applications (requiring higher than necessary privileges, or just plain buggy).
I'm hoping Longhorn will _default_ to a less-privileged user and that will "encourage" developers to start writing their software correctly (since ~10+ years of just waiting for them to do the right thing hasn't) - but this isn't any sort of low-level design or even implementation change, it's just a better default configuration (no different than enabling the firewall by default).
I've been using NT on my desktop (and to a lesser degree as a server) for nearly a decade now - not one virus, not one trojan, not one piece of spyware has ever managed to get itself onto any of my systems, and all I've ever used is common sense and the capabilities of the stock OS distribution (run as a low privilege user, don't run code from dodgy sources, avoid buggy software - ie: exactly the same techniques I use on unix and every other multiuser OS).
Re:Looks like they've got their focus...
on
Longhorn Preview
·
· Score: 1
Maybe the "new type of restricted user account" to start?
The only thing stopping everyone doing this now is poorly written applications. I personally have been running NT with a restricted account for nearly a decade now.
Re:Looks like they've got their focus...
on
Longhorn Preview
·
· Score: 1
Soooooo, little development with respect to security [...]
Given the vast bulk of security problems lie with applications and users, what else did you expect ?
Apparently desire to prevent a user from modifying a.plist file was not on Apple's mind, when they made the decision to store it in a binary form.
Apparently desire to stop the users from editing the registry was not on Microsoft's mind, either, when they provided not one, but two tools for doing so with the OS.
*Dicouraging* ? Certainly - as they should. Prevention ? No.
The difference is it is still not concentrated in one big file, which, if damaged, might render your system completely unuseable.
The probability of this happening/without deliberate action by the ignorant end user/, or some sort of hardware failure, is extremely small. About as likely as all those wonderful little files getting wiped out by some filesystem corruption.
That's all I'm talking about.
You're talking about an apples and oranges comparison - a rogue application that writes data in non-standard places in the registry vs an application that is properly behaved and keeps is config files in the proper place.
How does your method deal with a rogue application that writes files in random places in the directory structure ? It's not like the OS keeps track.
Instead, now to completely remove all traces of an application from your system, you have to spend a good deal of time and effort, searching for whatever keys it could have created in the registry, manually, and eradicating them, like pests. Poor and stupid design, IMHO.
The situation is _identical_ if an application storing its configuration in files behaves similarly. Your criticisms apply to application usage of the Registry, not the Registry itself.
Ask yourself what the process is for changing the IP address of a windows machine.
Trivial.
Then look at the process for changing the IP address for a Linux/Unix/Mac machine.
Which one ? While changing an IP on MacOS is fairly consistent across versions, changing an IP address on unix machines is very OS-dependent.
If your criteria for determining whether or not IP is "tacked on" is how difficult it is to change the machine's IP (ignoring for a second how fundamentally stupid that metric is), then IP is a hell of a lot more "tacked on" to the typical unix than Windows.
Look at the process for adding an IP printer vs. that of a NetBIOS printer.
I think you need to review the OSI model.
There are lots of examples, Windows just wasn't built with IP in mind - that's not the end of the world, but it shows through in the complexity of the interface.
O_o
You're seriously trying to suggest that a few obvious mouse clicks in Control Panel (or System Preferences) is "complex" compared with manually editing obscure (and inconsistent) text files ?
In Mac OS X user _DON'T HAVE TO_ edit any text files, but he can, if he needed to (the files are all in XML, and there's a pretty nice editor for them as well).
I never suggested OS X users had to edit text files. Also, perhaps you've noticed those xml text files are now compiled into binary files in Tiger ? Why do you think that is ?
Apple gets it - users SHOULD NOT BE directly editing configuration data, no matter how it is stored. Vast chunks of the unix and linux community haven't figured that out yet.
If a program can only store its preferences in its own subdirectory - then you don't even need to track what did it to there: once you want to delete it - delete the whole subdirectory, with whatever is stored inside. No cluttering of the system registry with all this garbage.
Just like you can delete registry keys, you mean ?
At some point I discovered that about 13 gigabytes of disk space on my hard drive were gone. [...]
And this proves my point. GP asks for some form of automated "revision control". I point out that a primitive form already exists, then you say it sucks because it uses disk space.
This is called wanting to have your cake, and eat it, too. Either you get a revision system and sacrifice some disk space to it, or you don't get it.
The subjects are clearly related: windows networking configurations are profoundly non-obvious.
I think you'll need to expand on that a bit.
In a lot of ways, it's similar to pre OSX Mac networking, which is to say that it's non-IP based, with IP grafted on.
Which part of IP do you think is "grafted on" and why ?
That there is one task which requires fewer clicks in Windows is not surprising, but let me give you the corresponding Mac example: try turning sharing OFF if someone has enabled some random folders...
It's in Control Panel. Actually, there are numerous ways of accessing the information, but that is the most obvious.
Slashdot Mirror
Not to detract from the rest of your argument, which I agree with, but most people can't type faster than (or even close to as fast as) they can write - they'll write something down rather than type it into notepad.
I think you'll find the ~5x improvement in bus bandwidth will give a quite healthy boost in real-life performance, even if the clock speed remains the same.
Minor changes /in Windows/. Applications may need major reworking - but that's not Microsoft's fault and there's nothing they can do about it.
What I'm trying to say here, is that everyone goes on about how Windows needs to be "redesigned" or "rewritten from the ground up" to be "secure" when all it really needs are a few default settings tweaked. The problem for Microsoft is that tweaking these defaults is potentially going to piss off a lot of people.
In fact, the only time I log on to her machine as admin is to use Windows Update or to install software.
If you're not already, you should really use "Run As" for this sort of thing.
Stop moving those goal posts around. Your original claim was that Windows was not "designed with IP in mind" based on the rather shaky assertion that all IP-related configuration was "complex". This has since devolved to basing your far-reaching criticism solely on the UI for setting up IP printers (a relatively uncommon task, even in networked environments).
I hope you realise how silly this line of reasoning is. It's like saying "Linux wasn't built with SCSI in mind" because fdisk is hard to use.
In any event, my comment on the OSI model was because you are trying to compare something at the network layer (IP) with something at the session layer (NetBIOS). NetBIOS runs *on top of IP* (or NetBEUI, or IPX, or probably anything else you want it to).
These links may be of assistance.
Windows BY DEFAULT prefers to use NetBIOS to IP.
Assuming you _really_ mean the old, unroutable NetBEUI protocol (that operates at the same level as IP), and not NetBIOS, it hasn't since Windows 98 (and even then, I seem to recall Windows 98 defaults to NetBIOS over IP - but it's been a long time since I've installed Windows 98).
NT variants of Windows have preferred NetBIOS over IP (as opposed to NetBIOS over NetBEUI, which is what I think you're talking about) since at _least_ NT4 (probably even earlier, but it's been a very long time since I've installed earlier versions of NT).
However, all this is irrelevant - Windows is no less (or more) "designed to use IP" than OS X is "designed to use Firewire". IP is simply a minor implementation detail of getting network data from machine A to machine B, just like Firewire is a minor implementation detail of how to connect a hard disk to a computer.
The ad-hoc filesharing which was mentioned earlier in the thread (right click, share this folder): what is that based on?
SMB (or CIFS, depending on what you want to call it). Again, complete independent of the network layer - it can run over IP, NetBEUI, IPX/SPX, etc (just like AFP on MacOS).
However, consider the paradigm: in Unix, EVERYTHING is manually editing obscure text files.
Ah, but some are more obscure than others. By your reasoning, some Linux distributions aren't "designed with IP in mind" and others are, despite being the same OS, merely because they happen to have a more complex UI for IP configuration.
Would you say that windows 95 was "built with IP in mind," or was IP "grafted on"?
Ditto for NT.
I would say it's completely and utterly irrelevant. Both of them were built to be network OSes. It's like asserting an OS was "built with SCSI in mind" or "built with AGP in mind" or "built with SDRAM in mind" whereas others had it "grafted on".
Again, I repeat my claim that IP on windows is in the Mac OS 8 realm: it works, but it's clunky, and it was treated as an afterthought when the OS was written.
And I repeat that basing your criticisms of low-level OS design on one example of clunky *user interface* is idiotic.
It's worth noting that the only platform that "requires driver source code" is Linux - everyone else's is stable enough such that hardware vendors can write their own drivers and not have to worry about the next sub-minor kernel revision breaking them.
Amazing how quick the battlecry goes from "users should have choice" to "users should use linux"...
Some programming would qualify as engineering - carefully specified, carefully implemented, deeply tested, task-specific.
Some programming would qualify as hobby - the people who try to roll together [some program] with [some wierd constraint] just for the hell of it.
I can't say I'd consider any sort of programming "art".
But - as I said - there's nothing stopping them from running as a regular user all the time either, except poorly written apps.
In this case, it's the operating system.
It's a minor default configuration detail. Hardly anything that requires major changes.
This is irrelevant to anyone not already indoctrinated into unix.
- All the features that would sell an OS upgrade have been pulled to meet the release date.
Much like the original OS X release, really...
Oddly enough, this doesn't seem to have slowed their penchant for clicking on "get your free b00bies here" attachments and installing Bonzai Buddy...
For some reason I'd be willing to bet they'll have an "update all selected files with [metadata]" in a file properties box.
Windows - like Linux and OS X - is only as insecure as you choose to make it.
That's because out in the real world, very few people care about rebooting (most of them shut the machine down every night anyway).
Actually, if you're an "Admin" user, you have write access (without any password prompt) to quite a bit of the system - probably everything in /Applications, for example (unless you've gone through an manually twiddled file permissions).
A virus could propogate quite nicely through an OS X system - assuming it could get itself executed by an "Admin" user (very high possibility) - simply by infecting nearly every application installed on the machine. For example (and I'm speaking hypothetically without fully researching the feasibility) I suspect it would be possible to have an Admin-run virus infect Installer.app and then use that as an avenue to achieve root privileges the next time some app's install procedure pops up the password box.
No, it's because different people have different tastes.
Depends on whether Apple has a patent on their "translucent kernel dump" screen.
How ? They don't write every application. They can't _make_ users not do stupid things.
The firewall in SP2 was a start, but how about making something that stops the OS being taken over after one wrong click on a message box in IE?
You can do that now - just don't run IE as an Administrator.
There are so many people with spyware/virus problems, including reasonably experienced users, that at some point Microsoft has to start taking the blame.
There are a handful of viruses that actually exploit OS vulnerabilities and bugs (most - if not all - of which have been fixed) everything else comes back to either the user (deliberately running malicious code, running with higher than necessary privileges) and applications (requiring higher than necessary privileges, or just plain buggy).
I'm hoping Longhorn will _default_ to a less-privileged user and that will "encourage" developers to start writing their software correctly (since ~10+ years of just waiting for them to do the right thing hasn't) - but this isn't any sort of low-level design or even implementation change, it's just a better default configuration (no different than enabling the firewall by default).
I've been using NT on my desktop (and to a lesser degree as a server) for nearly a decade now - not one virus, not one trojan, not one piece of spyware has ever managed to get itself onto any of my systems, and all I've ever used is common sense and the capabilities of the stock OS distribution (run as a low privilege user, don't run code from dodgy sources, avoid buggy software - ie: exactly the same techniques I use on unix and every other multiuser OS).
The only thing stopping everyone doing this now is poorly written applications. I personally have been running NT with a restricted account for nearly a decade now.
Given the vast bulk of security problems lie with applications and users, what else did you expect ?
Apparently desire to stop the users from editing the registry was not on Microsoft's mind, either, when they provided not one, but two tools for doing so with the OS.
*Dicouraging* ? Certainly - as they should. Prevention ? No.
The difference is it is still not concentrated in one big file, which, if damaged, might render your system completely unuseable.
The probability of this happening /without deliberate action by the ignorant end user/, or some sort of hardware failure, is extremely small. About as likely as all those wonderful little files getting wiped out by some filesystem corruption.
That's all I'm talking about.
You're talking about an apples and oranges comparison - a rogue application that writes data in non-standard places in the registry vs an application that is properly behaved and keeps is config files in the proper place.
How does your method deal with a rogue application that writes files in random places in the directory structure ? It's not like the OS keeps track.
Instead, now to completely remove all traces of an application from your system, you have to spend a good deal of time and effort, searching for whatever keys it could have created in the registry, manually, and eradicating them, like pests. Poor and stupid design, IMHO.
The situation is _identical_ if an application storing its configuration in files behaves similarly. Your criticisms apply to application usage of the Registry, not the Registry itself.
Trivial.
Then look at the process for changing the IP address for a Linux/Unix/Mac machine.
Which one ? While changing an IP on MacOS is fairly consistent across versions, changing an IP address on unix machines is very OS-dependent.
If your criteria for determining whether or not IP is "tacked on" is how difficult it is to change the machine's IP (ignoring for a second how fundamentally stupid that metric is), then IP is a hell of a lot more "tacked on" to the typical unix than Windows.
Look at the process for adding an IP printer vs. that of a NetBIOS printer.
I think you need to review the OSI model.
There are lots of examples, Windows just wasn't built with IP in mind - that's not the end of the world, but it shows through in the complexity of the interface.
O_o
You're seriously trying to suggest that a few obvious mouse clicks in Control Panel (or System Preferences) is "complex" compared with manually editing obscure (and inconsistent) text files ?
So you're trying to say that a 100GB library of MP3s - with no original media even having been purchased - is not a copyright violation ?
You'll have to pardon my scepticism.
I never suggested OS X users had to edit text files. Also, perhaps you've noticed those xml text files are now compiled into binary files in Tiger ? Why do you think that is ?
Apple gets it - users SHOULD NOT BE directly editing configuration data, no matter how it is stored. Vast chunks of the unix and linux community haven't figured that out yet.
If a program can only store its preferences in its own subdirectory - then you don't even need to track what did it to there: once you want to delete it - delete the whole subdirectory, with whatever is stored inside. No cluttering of the system registry with all this garbage.
Just like you can delete registry keys, you mean ?
At some point I discovered that about 13 gigabytes of disk space on my hard drive were gone. [...]
And this proves my point. GP asks for some form of automated "revision control". I point out that a primitive form already exists, then you say it sucks because it uses disk space.
This is called wanting to have your cake, and eat it, too. Either you get a revision system and sacrifice some disk space to it, or you don't get it.
I think you'll need to expand on that a bit.
In a lot of ways, it's similar to pre OSX Mac networking, which is to say that it's non-IP based, with IP grafted on.
Which part of IP do you think is "grafted on" and why ?
That there is one task which requires fewer clicks in Windows is not surprising, but let me give you the corresponding Mac example: try turning sharing OFF if someone has enabled some random folders...
It's in Control Panel. Actually, there are numerous ways of accessing the information, but that is the most obvious.