And in turn, CERT's vulnerability count for apache can demonstrate this statement is simply false.
And to qoute Shakespeare, "Even the devil can cite scripture for his purpose": if you want to fairly compare this to IIS's problem count, you should include an application scripting environment, as IIS includes ASP. Let's say PHP, since it seems to be the most popular; we get this count.
This is going to be controversial in some peoples' eyes, but you can "thank" the popularity of Windows for the cheapness and ubiquity of x86 hardware.
Of course, this also prevents more creative diversity in hardware, but do you want diversity in hardware or in software? It's an easy choice for me: modifying source is easier than trying to swap out a ROM...
MSI is a set of APIs exposed by a Windows Service; it is not actually a package per-se. These APIs are published, and public, so any third party install-packager software can generate MSI files; InstallShield & Wise to name two.
Actually, in Windows, virtual desktop support and fast user switching are two separate things. Virtual desktops are on a per-user basis -- 4 per user with the ms vdm -- and don't involve another logon procedure, so yep, they're a visual hack. OTOH, user switching/remote desktop/remote assistance and terminal server are all variations on the multi-user aspect, e.g. they're based around actual multiple logon sessions.
Very, very interesting! However, this may have happened because you created the "new" community using the same Passport. Have you tried creating communities with predictable names to see if you can ressurrect someone else's content? I seriously doubt it.
Sounds like the whole system is based on a FAT metaphor: your account [directory entry] was marked deleted but the content [raw sector content] was untouched. One big flaw though, reallocated sectors' content does'nt appear in the body of the file. I wonder if this was done for performance issues, except there's a glitch where it should be "quickformatting" the area?
Looking out for new/modifed files isn't always going to help you unfortunately. Badly written application installers will stomp on common DLLs, overwriting them with their own particular version. Sometimes they'll just upgrade the common DLL with a later -- and mostly compatible -- version. If you go just looking to remove the files that have been "touched" after the install, you run the risk of removing a DLL that was previously in use by other applications. Welcome to what is affectionately known as "DLL Hell".
The real culprits here are the crappy installers. The sooner all Windows apps are installed using MSI (microsoft installer) services, the easier it will be to audit and rollback, and monitor what is going on. MSI's scope is enormous: it is fully transactional; it audits/logs everything, and it supports every option you could wish for. However, because of this, it is inherently complex and not everyone is using the API: it's gone through 3 versions already.
Once Windows is built entirely on a JIT'ed.NET subsystem (hand me my shotgun Jeeves: there's another flock of pigs overhead), all DLLs will be able to sit in side-by-side more whereby multiple versions can exist; however, this is a long ways away.
Ah come on; you can do better than that weak offering. W
"formatting C drive, yes or no?", that sounds much more like a CLI than a GUI prompt. Any GUI I've ever used will say a lot more than that, more like:
"This will erase all data from your hard drive; are you sure you want to continue?"
Ok, you could then argue that if the user has no clue what a hard drive is, they might click ok. But the point is, if same user knew what a hard drive was, he wouldn't continue (talking about the GUI i/f here); However, if the same user sat at a command line, he wouldn't even get that far.
True, this is a vulnerablility. But this is not an implicit design flaw that is to blame because there is a concept of zones (read: domains) with associated security policies within the security model.
The problem is that the URL after the bang is being handled in the security context (read: zone) of the URL before the bang. It looks more like the parser for the URL moniker is badly written, e.g. it's an implementation problem, rather than a concrete design flaw? No? It's not a windows archtitecture flaw, but rather some sloppy programming on the part of whomever wrote the moniker handler?
>It would be more intuitive, but it is currently implemented badly.
Do you mean the *nix shell? Not following you here.
>For example, in Windows when a piece of software doesn't show up in the Start->Programs menu, it becomes necessary to delve into C:\program files\company name\application name.
To be pedantic, this is the choice of the software developers, not Windows. I would agree that convention seems to be lead this way at least 50% of the time. I mean, coming back to *nix, sometimes stuff is put in/usr/local/bin/ other times in/opt/ or/usr/bin/. So, if you try to run a command (given that you somehow now what it is to begin with), and it's not in your path, you must also delve. (yes, locate/updatedb, but this is analagous to windowskey+f in Windows).
> Why is there a quick launch menu to the left of the open windows
Eh, for a quick launch of a new process? No?
> and a running application tray which includes launchers to the right?
This is not an area for launching new processes -- although some things do use it as such, e.g winzip -- but rather for interacting with running processes that don't warrant a full slot in the taskbar.
> Why are both of these to the right of the start menu, which includes no less than two different places for application icons?
Can you expand on this point please?
> And why oh why are we not supposed to have icons on the desktop?
Why not? I have plenty on my desktop?
> What about launching in explorer? Without including the run menu, command-key launchers, or other non-graphical ways of launching applications, you already have 6 different places to look for an icon,
> and the icon might be in any or none of those. Add in that Windows is now a multi-user environment masquerading as a single-user, meaning that the icon might not be there at all in your user, and you see why so many people complain about icons being lost.
A fair point, sometimes people install software for "me" rather than "all users". Anyhow, if you're looking for an icon that you cannot find, how do you know it should be there in the first place? Have you ever found dangling symlinks? No? Well, lucky you.
> The engine communicating with the deflector shield I feel is an apt analogy, because the command line provides a simple way for applications to communicate.
Sure, if you need applications to communicate in that way. One thing Windows has over Gnome/KDE is a unified clipboard.
> While graphical user interfaces have interprocess communication, the process is far less elegant, expedient, and easy to code, and fits a far less general case.
This is hardly relevent; the kind of data exchanged through a CLI pipe is completely different that that is exchanged through IPC, as are the programming models that use it.
> Why is it when attacking the CLI people resort to showing a confused new user?
Who admires the ultra-cool hacker in the movies, completing unimaginably difficult tasks with him/her sitting at a command prompt, reems of text shooting up the terminal; who didn't notice the original Terminator's HUD spat out lines of Cobol; who was impressed when that girl in Jurassic Park spouted "This is unix!"; etc etc? I'll concede that the CLI isn't as difficult as it seems once you get stuck in, but the point is, it's unapproachable without coaching. You can grab a mouse and start clicking things, but faced with:
~#
Wtf do you do?
> but once acquainted with it (like the bridge of the enterprise should be) it is more powerful than a GUI for many tasks.
I agree 100%. I never argued against this. My point is that the learning curve is way, way higher than a GUI, for the average joe.
Hmmm. It's very interesting how that the most popular attack on GUIs is by playing up the whole "I don't know where the icon is,"
e.g. Worf : "Would that be in the engine program or the deflector shield program?"
How is this any different from not knowing the path of a command (not all commands are in $PATH), and more to the point, not knowing the syntax? (yes, -h, -?; but this is countered by Windows' F1 key for example).
Answer: there is no difference. But, people inexperienced with computers will pick up on visual clues given with colour, shape and the plain pictogram metaphors employed (sometimes good, sometimes bad).
How intuitive is ~$ ?
I am not a GUI zealot, by any means, and I use Windows (2000+) and Linux (Gentoo + RH) daily.
> there are students in my classes who couldn't use a terminal to save their lives or work remotely without a GUI. They just don't understand the system commands
Do you perceive this as a problem? What do you think 30+ years of GUI development has been for -- isn't this statement a proof of its [GUIs] success (or, gasp, is it proof of MS Windows' success?) What am I trying to say here?
Not trying to be difficult or anything, but I think this kind of comment is redundant. Don't get me wrong, I'm perfectly comfortable with command line tools, but hey, it's the 21st century. Imagine if the bridge of the Enterprise was commanded by a load of people typing 100-char length phrases into the computers:
Picard: "Worf! Red Alert! Shields Up!"
Worf: "Yes sir!"
~$ su - incorrect password *damn* ~$ su - ~# cd/bin/shields/ ~# shelds up 'shelds' command not found *damn*
Picard: "Worf?"
*noise of klingon disrupter ripping through hull*
Anyhow, you get my meaning. Cmdline is great for nostalgia most of the time, and sometimes you can't do without it, but this is a failing of the GUI in most cases, IMO.
Funny? Mod parent insightful! This is closer to the truth than most realise.
As much is it one of the great pluses of the Open Source movement to be able to branch like this, in reality, it's another truckload of headaches for compatilibity issues in the future. The X server has been one of the only invariables recently in the layering of our systems; now we have just witnessed another variable which doubles the possible combinations (read: problems) from the X server layer up. For those of us who want linux on the desktop (read: useable by noobs), this spells ever more woe. Ah well, at least sysadm salaries might increase out of this one.
Yes, you are correct. However, putting the drivers into "ring 0" as you say, has the unfortunate side effect of letting badly written drivers -- by 3rd party graphics card vendors -- bring down the system.
But, I think saying "microsoft thought it cool" is a bit of a simplification; NT was also trying to target the graphics workstation market also, as well as the home market. Having slow user space graphics drivers was not good for this, so putting them in kernel space was the only answer. Hence even with the late addition of DGI with xfree, Windows still dominates the gaming market.
Don't make the usual mistake of comparing NT based releases with 95/98 which you imply with your talk of stability. The latter releases are not worthy of mentioning in the same sentence as Operating System. They are just toys.
And as for your parting statement, "I wouldn't call it a microkernel," what is your definition of a Microkernel then?
My God, what a mindless mob of moderators we have today.
I've got mod points now, but rather than pointlessly mod down the parent, I've eschewed them to say this: How in the name of Linus's bumcheeks is reiterating business common sense -- try to dominate the market with your product -- insightful?
Do you not think that market dominance is not an appropriate goal for Linux? Do you think that the principal designers of NT are only interested in market control? You can't put together a operating system with marketing fiends using Powerpoint? (well, maybe windows 95 was a result of that).
Anyone care to back me up on this? Am I completely deluded?
Slashdot Mirror
And in turn, CERT's vulnerability count for apache can demonstrate this statement is simply false.
And to qoute Shakespeare, "Even the devil can cite scripture for his purpose": if you want to fairly compare this to IIS's problem count, you should include an application scripting environment, as IIS includes ASP. Let's say PHP, since it seems to be the most popular; we get this count.
Quite close, aren't they?
- Oisin
At least I can get an all-over "CRT tan" now; this will get to those hard to reach spots under the chin.
Well said! Mod parent +5 "shut those moaning ingrates right up"
- Oisin
This is going to be controversial in some peoples' eyes, but you can "thank" the popularity of Windows for the cheapness and ubiquity of x86 hardware.
Of course, this also prevents more creative diversity in hardware, but do you want diversity in hardware or in software? It's an easy choice for me: modifying source is easier than trying to swap out a ROM...
- Oisin
MSI is a set of APIs exposed by a Windows Service; it is not actually a package per-se. These APIs are published, and public, so any third party install-packager software can generate MSI files; InstallShield & Wise to name two.
- Oisin
Actually, in Windows, virtual desktop support and fast user switching are two separate things. Virtual desktops are on a per-user basis -- 4 per user with the ms vdm -- and don't involve another logon procedure, so yep, they're a visual hack. OTOH, user switching/remote desktop/remote assistance and terminal server are all variations on the multi-user aspect, e.g. they're based around actual multiple logon sessions.
- Oisin
- Oisin
With the Virtual Desktop Manager, four. 3rd party managers would probably give you more.
- Oisin
For someone called "ObviousGuy", you sure do come up with some obscure metaphors.
Buns, pickles and meat indeed. Pah.
- Oisin
Unlike Fisher-Price toys, Win XP does not have to stay green and chunky. You can apply your own themes. Not just colours, but complete skins.
- Oisin
Well, to be fair, TWM (or Tabbed Window Manager) doesn't have anything to do with Linux. In fact, it predates Linux by at least 5 years.
- Oisin
Very, very interesting! However, this may have happened because you created the "new" community using the same Passport. Have you tried creating communities with predictable names to see if you can ressurrect someone else's content? I seriously doubt it.
Sounds like the whole system is based on a FAT metaphor: your account [directory entry] was marked deleted but the content [raw sector content] was untouched. One big flaw though, reallocated sectors' content does'nt appear in the body of the file. I wonder if this was done for performance issues, except there's a glitch where it should be "quickformatting" the area?
- Oisin
Looking out for new/modifed files isn't always going to help you unfortunately. Badly written application installers will stomp on common DLLs, overwriting them with their own particular version. Sometimes they'll just upgrade the common DLL with a later -- and mostly compatible -- version. If you go just looking to remove the files that have been "touched" after the install, you run the risk of removing a DLL that was previously in use by other applications. Welcome to what is affectionately known as "DLL Hell".
.NET subsystem (hand me my shotgun Jeeves: there's another flock of pigs overhead), all DLLs will be able to sit in side-by-side more whereby multiple versions can exist; however, this is a long ways away.
The real culprits here are the crappy installers. The sooner all Windows apps are installed using MSI (microsoft installer) services, the easier it will be to audit and rollback, and monitor what is going on. MSI's scope is enormous: it is fully transactional; it audits/logs everything, and it supports every option you could wish for. However, because of this, it is inherently complex and not everyone is using the API: it's gone through 3 versions already.
Once Windows is built entirely on a JIT'ed
- Oisin
Ah come on; you can do better than that weak offering. W
"formatting C drive, yes or no?", that sounds much more like a CLI than a GUI prompt. Any GUI I've ever used will say a lot more than that, more like:
"This will erase all data from your hard drive; are you sure you want to continue?"
Ok, you could then argue that if the user has no clue what a hard drive is, they might click ok. But the point is, if same user knew what a hard drive was, he wouldn't continue (talking about the GUI i/f here); However, if the same user sat at a command line, he wouldn't even get that far.
- Oisin
True, this is a vulnerablility. But this is not an implicit design flaw that is to blame because there is a concept of zones (read: domains) with associated security policies within the security model.
w .e xample.com//exploit.chm::exploit.html
E.g.
ms-its:mhtml:file://C:\nosuchfile.mht!http://ww
The problem is that the URL after the bang is being handled in the security context (read: zone) of the URL before the bang. It looks more like the parser for the URL moniker is badly written, e.g. it's an implementation problem, rather than a concrete design flaw? No? It's not a windows archtitecture flaw, but rather some sloppy programming on the part of whomever wrote the moniker handler?
Just my 2 cents.
- Oisin
Speaking of Boy Bands though; couple these robo-gimps with a couple of n*sync masks and some of that Vocaloid magic.
Only thing is now, who shags the groupies?
>It would be more intuitive, but it is currently implemented badly.
/usr/local/bin/ other times in /opt/ or /usr/bin/. So, if you try to run a command (given that you somehow now what it is to begin with), and it's not in your path, you must also delve. (yes, locate/updatedb, but this is analagous to windowskey+f in Windows).
Do you mean the *nix shell? Not following you here.
>For example, in Windows when a piece of software doesn't show up in the Start->Programs menu, it becomes necessary to delve into C:\program files\company name\application name.
To be pedantic, this is the choice of the software developers, not Windows. I would agree that convention seems to be lead this way at least 50% of the time. I mean, coming back to *nix, sometimes stuff is put in
> Why is there a quick launch menu to the left of the open windows
Eh, for a quick launch of a new process? No?
> and a running application tray which includes launchers to the right?
This is not an area for launching new processes -- although some things do use it as such, e.g winzip -- but rather for interacting with running processes that don't warrant a full slot in the taskbar.
> Why are both of these to the right of the start menu, which includes no less than two different places for application icons?
Can you expand on this point please?
> And why oh why are we not supposed to have icons on the desktop?
Why not? I have plenty on my desktop?
> What about launching in explorer? Without including the run menu, command-key launchers, or other non-graphical ways of launching applications, you already have 6 different places to look for an icon,
> and the icon might be in any or none of those. Add in that Windows is now a multi-user environment masquerading as a single-user, meaning that the icon might not be there at all in your user, and you see why so many people complain about icons being lost.
A fair point, sometimes people install software for "me" rather than "all users". Anyhow, if you're looking for an icon that you cannot find, how do you know it should be there in the first place? Have you ever found dangling symlinks? No? Well, lucky you.
> The engine communicating with the deflector shield I feel is an apt analogy, because the command line provides a simple way for applications to communicate.
Sure, if you need applications to communicate in that way. One thing Windows has over Gnome/KDE is a unified clipboard.
> While graphical user interfaces have interprocess communication, the process is far less elegant, expedient, and easy to code, and fits a far less general case.
This is hardly relevent; the kind of data exchanged through a CLI pipe is completely different that that is exchanged through IPC, as are the programming models that use it.
> Why is it when attacking the CLI people resort to showing a confused new user?
Who admires the ultra-cool hacker in the movies, completing unimaginably difficult tasks with him/her sitting at a command prompt, reems of text shooting up the terminal; who didn't notice the original Terminator's HUD spat out lines of Cobol; who was impressed when that girl in Jurassic Park spouted "This is unix!"; etc etc? I'll concede that the CLI isn't as difficult as it seems once you get stuck in, but the point is, it's unapproachable without coaching. You can grab a mouse and start clicking things, but faced with:
~#
Wtf do you do?
> but once acquainted with it (like the bridge of the enterprise should be) it is more powerful than a GUI for many tasks.
I agree 100%. I never argued against this. My point is that the learning curve is way, way higher than a GUI, for the average joe.
Anyhow, nice to have dialog.
- Oisin
Hmmm. It's very interesting how that the most popular attack on GUIs is by playing up the whole "I don't know where the icon is,"
e.g.
Worf
: "Would that be in the engine program or the deflector shield program?"
How is this any different from not knowing the path of a command (not all commands are in $PATH), and more to the point, not knowing the syntax? (yes, -h, -?; but this is countered by Windows' F1 key for example).
Answer: there is no difference. But, people inexperienced with computers will pick up on visual clues given with colour, shape and the plain pictogram metaphors employed (sometimes good, sometimes bad).
How intuitive is ~$ ?
I am not a GUI zealot, by any means, and I use Windows (2000+) and Linux (Gentoo + RH) daily.
- Oisin
> there are students in my classes who couldn't use a terminal to save their lives or work remotely without a GUI. They just don't understand the system commands
/bin/shields/
Do you perceive this as a problem? What do you think 30+ years of GUI development has been for -- isn't this statement a proof of its [GUIs] success (or, gasp, is it proof of MS Windows' success?) What am I trying to say here?
Not trying to be difficult or anything, but I think this kind of comment is redundant. Don't get me wrong, I'm perfectly comfortable with command line tools, but hey, it's the 21st century. Imagine if the bridge of the Enterprise was commanded by a load of people typing 100-char length phrases into the computers:
Picard: "Worf! Red Alert! Shields Up!"
Worf: "Yes sir!"
~$ su -
incorrect password
*damn*
~$ su -
~# cd
~# shelds up
'shelds' command not found
*damn*
Picard: "Worf?"
*noise of klingon disrupter ripping through hull*
Anyhow, you get my meaning. Cmdline is great for nostalgia most of the time, and sometimes you can't do without it, but this is a failing of the GUI in most cases, IMO.
Thoughts?
- Oisin
> Could you imagine the kind of embarassing situations that would arise from being beamed out of the toilet?
ROTFLMAO!
Imagine: Spock appearing semi-naked in the transporter room, mid-tugging session.
Funny? Mod parent insightful! This is closer to the truth than most realise.
As much is it one of the great pluses of the Open Source movement to be able to branch like this, in reality, it's another truckload of headaches for compatilibity issues in the future. The X server has been one of the only invariables recently in the layering of our systems; now we have just witnessed another variable which doubles the possible combinations (read: problems) from the X server layer up. For those of us who want linux on the desktop (read: useable by noobs), this spells ever more woe. Ah well, at least sysadm salaries might increase out of this one.
My two cents.
- Oisin
You can't trust those statisticians though; 76% of statistics are just made up on the spot.
- Ois
Yes, you are correct. However, putting the drivers into "ring 0" as you say, has the unfortunate side effect of letting badly written drivers -- by 3rd party graphics card vendors -- bring down the system.
But, I think saying "microsoft thought it cool" is a bit of a simplification; NT was also trying to target the graphics workstation market also, as well as the home market. Having slow user space graphics drivers was not good for this, so putting them in kernel space was the only answer. Hence even with the late addition of DGI with xfree, Windows still dominates the gaming market.
Don't make the usual mistake of comparing NT based releases with 95/98 which you imply with your talk of stability. The latter releases are not worthy of mentioning in the same sentence as Operating System. They are just toys.
And as for your parting statement, "I wouldn't call it a microkernel," what is your definition of a Microkernel then?
- Oisin
Do you not think that market dominance is not an appropriate goal for Linux?
I didn't not mean to never not use (nor disuse) double negatives (nor inverse positives) like that, no, or maybe yes? Sorry.
- Oisin
FYI: NT is also microkernel based, 3.5/4/2000/XP and 2003.
- Oisin
My God, what a mindless mob of moderators we have today.
I've got mod points now, but rather than pointlessly mod down the parent, I've eschewed them to say this: How in the name of Linus's bumcheeks is reiterating business common sense -- try to dominate the market with your product -- insightful?
Do you not think that market dominance is not an appropriate goal for Linux? Do you think that the principal designers of NT are only interested in market control? You can't put together a operating system with marketing fiends using Powerpoint? (well, maybe windows 95 was a result of that).
Anyone care to back me up on this? Am I completely deluded?
- Oisin