You are missing the point about mounting. This is the big difference between windows and Unixes. Nothing is copyrighted and the file system is designed to be distributed. You don't need to push files at all. You don't log onto each machine. You change it at the mount point and it just is changed for all the clients.
So in other words there network config information comes in say/usr/networkadmin/etc and some files in/etc (like hosts) are symlinked into/usr/networkadmin/etc. And/usr/networkadmin/etc isn't on the the local machine it is a remote mount. It never gets "pushed" to the machines because it doesn't have to be.
It is a pity you are an AC because this is a good comment.
Explain why you think this is complex with different mount points, you configure a single computer and either
1) mount 2) copy directories
Remember you have clean separations. You need to install new apps that what package management is for. You need to change network configurations you push through a change to/etc.
One of the reasons Unix has all these mount points is so that various directories can be mounted remotely.
For example to control applications you can just have/usr be on a different partition which you control, not on their local machine at all. Or just/usr/etc. You can have files in/etc link off to files in say/usr/extrastuff/etc.
This is classically why/bin,/usr/bin,/usr/local/bin and/home/(username)/bin are 4 separate directories because they can be mounted 4 entirely different ways. You are trying to solve a problem Unixes don't have.
How many desktop machines actually need the/usr/ vs./usr/local/ distinction?
Actually quite a few. It comes up all the time on OSX. Apple installs to/usr while unixy software frequently installs to/usr/local. Oracle in their 10.1 release for Apple made heavy use of/usr/local as the default location for 4 shell scripts that configured the system properly for the Oracle layout.
On Linux I did it all the time./usr = stuff that came from the distribution ~/bin = stuff I wanted to use for myself that I I compiled or scripted myself/usr/local = stuff I wanted to make available to all users that I compiled or scripted myself
Going back even further I used it even more heavily on Sun workstations where I had/usr/local = stuff I wanted to make available on all my workstations/usr/local/local = stuff specific to this workstation/usr = stuff from Sun.
SunOS, AIX...
So my answer it has been useful to me on workstations since I started.
The example was wiping off information when you are done, which implies the system is multi-user. If it is single user then you are wiping the system for resale, fine wipe/etc.
Good point. Particularly in this day of virtualization./bin can be shared (and thus used within the virtualizer or across OSes while/usr/bin cannot be shared.
Not at all. Professional applications can assume things like background and vocabulary that are not available to general users. They can assume higher levels of interest and specialization. Intuitive interfaces seek to hide complexity, professional interfaces seek to organize complexity so that as much of it as possible can be displayed.
Does anybody honestly think that the traditional Unix filesystem heirarchy makes an ounce of sense in 2009
It doesn't make any sense when everything is on one partition. Gobolinux as a series of symlinks makes sense. Where there is a problem is in how directories are mounted. For example/usr/local/usr and/bin are often on different machines or have different maintenance schedules./home may be distributed across the enterprise or the collection of machines./opt can have all sorts of weird licensing issues which address how they can be shared and cross mounted.
You may want to distribute drive loads. Fragmentation results from having frequently altered files with infrequently altered files. That's why for example you want/usr and/var on separate partitions. Quite often you want to be able to wipe machines without altering the contents of/home. For example the whole problem with SSD wear could be solved by having different types of SSD for the/var partitions.
You don't want to be backing up/var/tmp.
Yes it makes a lot of sense and continues to. Now presenting it to the user via. symlinks in a more "logical way" is a good thing IMHO.
Or the content providers partner with device manufacturers and provide the content in a more limited way, which is what they are currently doing. From Microsoft's perspective what they see is:
Option 1: Provide a DRM solution to customers to allow them to do some of what they want Option 2: Provide no solution to customers
The goal of DRM in Vista is not to prevent play it is to prevent recording.
In other words BluRay DVD -> Player -> Screen is indistinguishable to the player from: BluRay DVD -> Player -> virtual device -> ripping software -> diskfile
However the OS can tell the difference.
And yes you can have the keys and lock and not be able to get in the door. That is the whole theory of encryption. I can allow you to perform a mathematical operation that I can reverse but you can't. Just read up on trusted computing.
Program X can't tell if it is sending a message to a virtual device or a real device. Only the OS can do that (sort of). And to really do it, you need hardware support as well.
Yes you need DRM in the OS and the in hardware if you want DRM to actually work.
I don't see a main post on this but Microsoft deserves kudos for fixing the standards problem. They get a lot of slack around here when they do something wrong they should get credit for doing something right.
I think ActiveX was great, at least as a proprietary solution. Particular in I.E. 4.5 when it was integrated with Active desktop and push technologies. You could do amazing things 10 years ago you still can't really do today. What it wasn't was secure.
NT had the security to handle this, but that wasn't turned on because they wanted to hide the differences from Win98 and not have two very different product lines. Further corporate IT wanted control.
At the time of NS4 there were a bunch of other browsers on the market like Konq. Had Microsoft worked with the other browsers timeline we could have had a standards body.
How about don't use autoupdate and just download the update files on to your Apple Server and just deploy them one after another. Takes about a minute to wrap them this way.
Actually a lot of sites went for IE6 compliance and didn't bother to make 2 versions. They didn't want to pay the extra 30% to have full standard compliance.
Yep you got it. There are stable products and there are unstable products. You want stability Z/OS and COBOL offer a great platform where all changes are discussed over the process of a decade.
No one ever said that would be the case with the internet in the mid to late 90's when these sites were created.
Maybe. What really drove it was the unresponsiveness of the corporate IT departments to user demand. The "rogue" IS problem. IT departments managed to reassert control via the "managed PC".
What's happening with SAS is that internal IT departments are getting competition from external IT departments. The end users have access to a wealth of applications even on their managed PCs. They have choice and departments that are anxious to provide services just like the late 80's - mid 90s.
Given the current managed environment I see the move internet apps as increasing freedom.
Slashdot Mirror
You are missing the point about mounting. This is the big difference between windows and Unixes. Nothing is copyrighted and the file system is designed to be distributed. You don't need to push files at all. You don't log onto each machine. You change it at the mount point and it just is changed for all the clients.
So in other words there network config information comes in say /usr/networkadmin/etc and some files in /etc (like hosts) are symlinked into /usr/networkadmin/etc. And /usr/networkadmin/etc isn't on the the local machine it is a remote mount. It never gets "pushed" to the machines because it doesn't have to be.
It is a pity you are an AC because this is a good comment.
Explain why you think this is complex with different mount points, you configure a single computer and either
1) mount
2) copy directories
Remember you have clean separations. You need to install new apps that what package management is for. You need to change network configurations you push through a change to /etc.
One of the reasons Unix has all these mount points is so that various directories can be mounted remotely.
For example to control applications you can just have /usr be on a different partition which you control, not on their local machine at all. Or just /usr/etc. You can have files in /etc link off to files in say /usr/extrastuff/etc.
This is classically why /bin, /usr/bin, /usr/local/bin and /home/(username)/bin are 4 separate directories because they can be mounted 4 entirely different ways. You are trying to solve a problem Unixes don't have.
How many desktop machines actually need the /usr/ vs. /usr/local/ distinction?
Actually quite a few. It comes up all the time on OSX. Apple installs to /usr while unixy software frequently installs to /usr/local. Oracle in their 10.1 release for Apple made heavy use of /usr/local as the default location for 4 shell scripts that configured the system properly for the Oracle layout.
On Linux I did it all the time. /usr = stuff that came from the distribution /usr/local = stuff I wanted to make available to all users that I compiled or scripted myself
~/bin = stuff I wanted to use for myself that I I compiled or scripted myself
Going back even further I used it even more heavily on Sun workstations where I had /usr/local = stuff I wanted to make available on all my workstations /usr/local/local = stuff specific to this workstation /usr = stuff from Sun.
SunOS, AIX...
So my answer it has been useful to me on workstations since I started.
The example was wiping off information when you are done, which implies the system is multi-user. If it is single user then you are wiping the system for resale, fine wipe /etc.
Good point. Particularly in this day of virtualization. /bin can be shared (and thus used within the virtualizer or across OSes while /usr/bin cannot be shared.
I use to go to websites and thus it should be deleted when I clear my private data.
Which will fail because you don't have permissions to edit /etc unless you are an administrative user, i.e. logged in as root. So what is the problem.
Not at all. Professional applications can assume things like background and vocabulary that are not available to general users. They can assume higher levels of interest and specialization. Intuitive interfaces seek to hide complexity, professional interfaces seek to organize complexity so that as much of it as possible can be displayed.
Both emacs and vi/vim have excellent documentation and have had it for decades.
Does anybody honestly think that the traditional Unix filesystem heirarchy makes an ounce of sense in 2009
It doesn't make any sense when everything is on one partition. Gobolinux as a series of symlinks makes sense. Where there is a problem is in how directories are mounted. For example /usr/local /usr and /bin are often on different machines or have different maintenance schedules. /home may be distributed across the enterprise or the collection of machines. /opt can have all sorts of weird licensing issues which address how they can be shared and cross mounted.
You may want to distribute drive loads. Fragmentation results from having frequently altered files with infrequently altered files. That's why for example you want /usr and /var on separate partitions. Quite often you want to be able to wipe machines without altering the contents of /home. For example the whole problem with SSD wear could be solved by having different types of SSD for the /var partitions.
You don't want to be backing up /var/tmp.
Yes it makes a lot of sense and continues to. Now presenting it to the user via. symlinks in a more "logical way" is a good thing IMHO.
What was your PhD on? Always like to hear about Haskell projects.
I'd say you earned your user name on that one. I wasn't even sure if you were being sarcastic.
Or the content providers partner with device manufacturers and provide the content in a more limited way, which is what they are currently doing. From Microsoft's perspective what they see is:
Option 1: Provide a DRM solution to customers to allow them to do some of what they want
Option 2: Provide no solution to customers
The goal of DRM in Vista is not to prevent play it is to prevent recording.
In other words
BluRay DVD -> Player -> Screen
is indistinguishable to the player from:
BluRay DVD -> Player -> virtual device -> ripping software -> diskfile
However the OS can tell the difference.
And yes you can have the keys and lock and not be able to get in the door. That is the whole theory of encryption. I can allow you to perform a mathematical operation that I can reverse but you can't. Just read up on trusted computing.
Program X can't tell if it is sending a message to a virtual device or a real device. Only the OS can do that (sort of). And to really do it, you need hardware support as well.
Yes you need DRM in the OS and the in hardware if you want DRM to actually work.
I'm one of them. I've been watching this. I'm thrilled that Microsoft is taking a stand for quality.
No, this Microsoft's problem. You can check for version number so if I.E. 6 do X otherwise do Y.
I don't see a main post on this but Microsoft deserves kudos for fixing the standards problem. They get a lot of slack around here when they do something wrong they should get credit for doing something right.
I think ActiveX was great, at least as a proprietary solution. Particular in I.E. 4.5 when it was integrated with Active desktop and push technologies. You could do amazing things 10 years ago you still can't really do today. What it wasn't was secure.
NT had the security to handle this, but that wasn't turned on because they wanted to hide the differences from Win98 and not have two very different product lines. Further corporate IT wanted control.
At the time of NS4 there were a bunch of other browsers on the market like Konq. Had Microsoft worked with the other browsers timeline we could have had a standards body.
How about don't use autoupdate and just download the update files on to your Apple Server and just deploy them one after another. Takes about a minute to wrap them this way.
What determines the right way of doing things again?
W3C is who determines the right way of doing things. Pragmatism determines the quick and easy way of doing things. This isn't a complex question.
I got no problem telling businessmen they got to pay twice to build something they built wrong on purpose.
Actually a lot of sites went for IE6 compliance and didn't bother to make 2 versions. They didn't want to pay the extra 30% to have full standard compliance.
Yep you got it. There are stable products and there are unstable products. You want stability Z/OS and COBOL offer a great platform where all changes are discussed over the process of a decade.
No one ever said that would be the case with the internet in the mid to late 90's when these sites were created.
Maybe. What really drove it was the unresponsiveness of the corporate IT departments to user demand. The "rogue" IS problem. IT departments managed to reassert control via the "managed PC".
What's happening with SAS is that internal IT departments are getting competition from external IT departments. The end users have access to a wealth of applications even on their managed PCs. They have choice and departments that are anxious to provide services just like the late 80's - mid 90s.
Given the current managed environment I see the move internet apps as increasing freedom.