RPM's advantage is that it's based on cpio, (a little more flexible and less legacy-bound than ar or tar), and like java and "jar", they use special metadata files in the archive to encode special behavior and allow for backwards/forwards compatible format upgrades. (It's fairly trivial to strip off the RPM header and extract the cpio archive)
Well, that's not much of an advantage.
The other is that RPM uses a berkeley DB as a backend instead of flat text files. This doesn't make it easier to work with, but it's interesting because you can do things like ask rpm difficult questions... (who depends on this package? what package owns this file? what files have changed? where is the documentation for such and such?) librpm and the perl and python wrappers exist to make your sysadmin life easier dealing with that without trying to script the rpm front-end (whose syntax leaves something to be desired...)
But there really isn't any other differences. They both use a back-end dependancy database, have auto-resolving front ends, specialized build tools, and while dpkg has "control" files, RPM has "spec" files.
I think RPM not having an equivalent for dselect or APT out of the gate doomed it in the minds of many who experienced RPM hell. (I did, and I emerged all the better: now I package software for my own machines!)
NMCI is evil. EDS is retarded. We were asked to look to see if we could help identify how to integrate the legacy and C&C systems with the NMCI network. I wouldn't touch that work with a 10 foot pole. It mystifies and annoys anyone working on it.
... that the emergence of a social health care system could be considered an evolutionary step? We create social systems and technological constructs to replace the natural evolutionary pressures that we have sidestepped. We create them _oursevles_. It's in our nature.
The really useful local vulnerabilities are the variety that exploit buffer overflows in system calls to either set capability bits or the effective UID of a process running as a local user. The really clever ones setup a tasklet that spawns a root shell after a random delay so you can't pinpoint the creation of the shell process with a system call. No DDOS, no generated logs of weird URLs, no audit trail generated... Stealth, bitches!
(This is why I keep linux and solaris boxes in convienent places at home and at work. Oh! Such and such is acting up? Well just throw the CD into machine ABC for me please... klicka-klacka-klik)
But what I was trying to express was that bundling RDP into XP and Server 2003 gave NT the first possibility of location independance (something taken for granted on other networked OSs) and that is _why_ its significant. Provided you don't have the option of using other, less crippled OSs.
It changes how you do things. You don't install applications on every end-terminal if you don't have to. You don't force people to have to use complex VPN solutions from home anymore.
It's the difference between an SSH-hosted pty and/dev/console. To an application, there isn't a difference. This is the important thing.
VNC and Timbuktu and GotomyPC and even BackOrifice were 2nd-class citizens... hacks that we had to live with. Some applications didn't work right over these links. But now have the tsclient.sys video driver (like pseudotty)... multiple window stations (like multiple login sessions), so applications work (even if they try to use DirectX or DirectShow) with graceful fallback, from anywhere, without extra software, in a non-proprietary protocol that is reasonably secure.
That is a first. Something that we should keep pushing Microsoft to do.
That would require the OS vendor to have, on file, a list of every binary of every version of piece of software anyone would care to run on the system. (Viruses patch programs, so simple checksums don't really help you there). The most you could do is identify specific pieces of stand-alone malware, which is dwarfed by many other types of attacks.
This is an essentially impossible task, and it would suck for developers. It also puts the burden on the OS vendor to decide good vs. bad software. It's a sometimes arbitrary decision that is best serviced by having multiple competing vendors for a end-user to choose from (one that protects a user from softwares most out-of-line with that users' interests). The OS vendor could also abuse this position to reduce competition (not promptly signing or labeling software it doesn't like as "bad").
Yeah, so no. I think the current state-of-the-art (code fingerprinting and multiple vendors) is probably a more scalable, less big-brotherish solution.
Any virus or trojan has two parts. The first is the route of infiltration, the second is the payload which has the ill effects.
OS vendors can take care of the 1st part, and they already do. In theory, with a perfect OS (and a perfect user), there is no need for an antivirus or antispyware application.
However users aren't perfect, and even with patched machines they double click hotmamas.jpg.exe or britneyspears.jpg.dmg and click yes yes yes... and then it's the job of tools like Symantec to identify and nullify the payloads (a vastly different problem then "securing the borders").
Virus scanners search for undesirable (yet valid) code. OS vendors would be hard pressed to categorize any code as such; they need to stay neutral. But they should at least make sure the OS is robust, so only what the user intends to run is what is actually run in the first place.
Not exactly. NT4 is Windows 4.0. 2000 is Windows 5.0 XP is 5.1 2003 is 5.2 XP SP2 and Server 2003 SP 1 are both 5.2
(using NTSwitch on XP gives you Whistler Server (2003 beta), on 2003 gives you XP SP1, on 2003 SP1 and XP SP2 switches between each other)
Among the Server and Workstation variants of each version number, all DLLs can be swapped about. The lesser variants are just missing DLLs and linked INFs in the installer that comes on the CD. You can't swap DLLs between revisions, stuff breaks. NT4 cannot become 2003. But you can certainly convert Workstation to Server and back. All ntswitch does is change two registry entries... this is what makes a system behave as intended. The missing files are just that... missing. You can drop them in and everything works fine. Nothing gets overwritten.
The exception is Datacenter/Cluster editions. The kernels on these editions are slightly different enough to make them unique. But there still are vastly many similarities.
>> The only feature I miss is remote desktop, and that's only of marginal utility.
That's the most important reason why to install Server 2003 or XP. Once you start using it, it changes the way you work with Windows machines.
I suggest trying to find a copy of Server 2000 so at least you get Terminal Services (with unlimited connections in Per-User mode!). If you're too poor to spring for it, or don't trust P2P, you should try to find NTSwitch.exe... and follow these instructions:
- Execute the NTSwitch Program (Backup your system first) following the instructions that it gives
- You MUST immediately afterward successfully install (any) Service Pack. It apparantly creates/restores some necessary registry entries.
- After Service Pack is installed REBOOT machine.
When you go into the START MENU>Settings>Control Panel>"Add Remove Programs" and click on "Add/Remove Windows Components" you will get a series of errors - it will tell you what files that are missing.
You will need to obtain these files either from an existing W2K Server installation or from the 2000 Server install CD.
Copy all.inf files to the Windows\Inf directory Copy all.dll files to the System32\Setup directory
If this is done correctly then when you run the Add/Remove component it will list (2) Terminal Services options
You will still need to have either a W2K Server or Advanced Server CD to actually install the remaining Terminal Server files (apart from the ones above), these are located in a compressed format on the \I386 directory (TSC.001) on the CD (about 14MB)
Once you verify that Terminal Services is running and installed, you can revert the machine to Professional (or keep it at Server if you find it useful). Seeing a 2K professional machine running multiple Terminal Services sessions without protest is a clear indication that the Server vs. Workstation distinction is only for market segmentation and maximizing profit, not any technical/support reason.
At least in my area, they've replaced the hardware on the street corner. We now have fiber coming into the junction box down the street, and copper going to each house. They've moved to using a cheaper, fewer-node headend a few hundred feet away, instead of having a large installation in the distribution facility. I imagine this headend works like a switch, not a hub, because I can't sniff anything.
And they've switched to randomly assigning DHCP addresses with non-overlapping subnets (very frustrating for in-house networking). Sigh.
This is coming from someone who already pays for cable TV. Streaming multicast video from local networks? It'd be like having my own satellite feed. CNN Pipeline and other current video-on-demand stuff is a weak attempt; unambitious and ultimately flawed in execution.
Anyone can blog from anywhere. There are RSS->blog gateways, and SMTP->RSS gateways.
At some point someone's going to get clever and collapse all these concepts into "message atoms". Descriptive text, along with tagged URLs and attachments that are treated as a unit with an author, publish date, keywords, "parent atom" for replies, etc.
Weblog, forum, RSS feed, email, XMPP (Jabber, Google Talk)... these are all just retrieval/display methods.
The future of blogging is when a standard gets created (similar to the SMTP MIME envelope standard or XMPP) that appropriately captures this concept and such that all such instances of it can be cast into the standard.
Then create gateways and display systems, database schemas, etc. that can handle these atoms and give us true independance from the medium and increased focus on the message.
The gconf configuration contains a lot of strings that applications use for help text and dialog boxes where options are displayed. Usually they are packaged with _every_ translation included. So the files can be huge if they have 100 copies of every string. Plus it's XML so it's wordy.
Well, I would argue that magtape is a much better medium for capturing audio than a record, because the record has the disadvantage of being a physical process (needle being dragged around in resin) while the tape recording has little system reactivity (it doesn't take much EMF to line up domains on a thin film of tape).
That's why analog recording sessions are mastered to tape. Not vinyl.
Vinyl just so happened to be easy to reproduce, ship, and store. They suck as a reproductive medium. While the channel capacity is potential very high, the SNR is high as well, and the creation process introduces some non-tonal distortions that are difficult to filter or compensate for on playback.
Give me a digital reproductive medium anyday. Lasts forever and a good downmix speaks louder than words.
You can convert Dolby AC3 from DVDs into 5.1 channel OGG Vorbis streams to make DVD rips but it doesn't save you much bandwidth since AC3 is only 192kbps to begin with (can you believe it?)
I have a really nice OGG of the THX intro sound in 5.1 which sounds _much better_ than the AC3 version because of the higher bitrate ceiling. If I can find it I'll reply with a post to a link.
I don't have the slightest hesitation in using it to produce software that I might end up selling to someone else. But I would never turn around and expect the same level of service that I would expect as a personal or hobbyist user. If I was using it while working for a corporation I'd ask for the commercial license and also push for yearly maintenance, just to cover our collective asses.
Slashdot Mirror
... you actually have to follow up on and read all the links to "verify" the article summaries. (Did they *really* mean that?)
No matter what zany thing someone suggests furries might do, you can find a depiction easily with a google search and site:vclart.net
You do realize that like every Latitude, Optiplex and PowerEdge works perfectly with RHEL 3 and 4 out of the box with all the hardware options?
RPM, dpkg, it's all the same.
RPM's advantage is that it's based on cpio, (a little more flexible and less legacy-bound than ar or tar), and like java and "jar", they use special metadata files in the archive to encode special behavior and allow for backwards/forwards compatible format upgrades. (It's fairly trivial to strip off the RPM header and extract the cpio archive)
Well, that's not much of an advantage.
The other is that RPM uses a berkeley DB as a backend instead of flat text files. This doesn't make it easier to work with, but it's interesting because you can do things like ask rpm difficult questions... (who depends on this package? what package owns this file? what files have changed? where is the documentation for such and such?) librpm and the perl and python wrappers exist to make your sysadmin life easier dealing with that without trying to script the rpm front-end (whose syntax leaves something to be desired...)
But there really isn't any other differences. They both use a back-end dependancy database, have auto-resolving front ends, specialized build tools, and while dpkg has "control" files, RPM has "spec" files.
I think RPM not having an equivalent for dselect or APT out of the gate doomed it in the minds of many who experienced RPM hell. (I did, and I emerged all the better: now I package software for my own machines!)
NMCI is evil. EDS is retarded.
We were asked to look to see if we could help identify how to integrate the legacy and C&C systems with the NMCI network.
I wouldn't touch that work with a 10 foot pole. It mystifies and annoys anyone working on it.
... that the emergence of a social health care system could be considered an evolutionary step? We create social systems and technological constructs to replace the natural evolutionary pressures that we have sidestepped. We create them _oursevles_. It's in our nature.
The really useful local vulnerabilities are the variety that exploit buffer overflows in system calls to either set capability bits or the effective UID of a process running as a local user. The really clever ones setup a tasklet that spawns a root shell after a random delay so you can't pinpoint the creation of the shell process with a system call.
No DDOS, no generated logs of weird URLs, no audit trail generated...
Stealth, bitches!
That's just it.
They base certain aspects of Springfield on many real-life Springfields and other American towns.
They wanted a town they could be "rivals" with.
So they looked at a map of Sprinfield IL and found the nearest large town nearby. Thus begat Shelbyville.
nt
... it's a testament to the speed of the Google code slingers.
As the old slashdotism proclaims: "Nothing to see here. Move along"
It's a nice anecdote.
/dev/console. To an application, there isn't a difference. This is the important thing.
... multiple window stations (like multiple login sessions), so applications work (even if they try to use DirectX or DirectShow) with graceful fallback, from anywhere, without extra software, in a non-proprietary protocol that is reasonably secure.
(This is why I keep linux and solaris boxes in convienent places at home and at work. Oh! Such and such is acting up? Well just throw the CD into machine ABC for me please... klicka-klacka-klik)
But what I was trying to express was that bundling RDP into XP and Server 2003 gave NT the first possibility of location independance (something taken for granted on other networked OSs) and that is _why_ its significant. Provided you don't have the option of using other, less crippled OSs.
It changes how you do things. You don't install applications on every end-terminal if you don't have to. You don't force people to have to use complex VPN solutions from home anymore.
It's the difference between an SSH-hosted pty and
VNC and Timbuktu and GotomyPC and even BackOrifice were 2nd-class citizens... hacks that we had to live with. Some applications didn't work right over these links.
But now have the tsclient.sys video driver (like pseudotty)
That is a first. Something that we should keep pushing Microsoft to do.
I think the red, green and blue stuff is malarky.
I say it's cyan, magenta and yellow!
That would require the OS vendor to have, on file, a list of every binary of every version of piece of software anyone would care to run on the system. (Viruses patch programs, so simple checksums don't really help you there). The most you could do is identify specific pieces of stand-alone malware, which is dwarfed by many other types of attacks.
This is an essentially impossible task, and it would suck for developers. It also puts the burden on the OS vendor to decide good vs. bad software. It's a sometimes arbitrary decision that is best serviced by having multiple competing vendors for a end-user to choose from (one that protects a user from softwares most out-of-line with that users' interests). The OS vendor could also abuse this position to reduce competition (not promptly signing or labeling software it doesn't like as "bad").
Yeah, so no. I think the current state-of-the-art (code fingerprinting and multiple vendors) is probably a more scalable, less big-brotherish solution.
Any virus or trojan has two parts. The first is the route of infiltration, the second is the payload which has the ill effects.
OS vendors can take care of the 1st part, and they already do. In theory, with a perfect OS (and a perfect user), there is no need for an antivirus or antispyware application.
However users aren't perfect, and even with patched machines they double click hotmamas.jpg.exe or britneyspears.jpg.dmg and click yes yes yes... and then it's the job of tools like Symantec to identify and nullify the payloads (a vastly different problem then "securing the borders").
Virus scanners search for undesirable (yet valid) code. OS vendors would be hard pressed to categorize any code as such; they need to stay neutral. But they should at least make sure the OS is robust, so only what the user intends to run is what is actually run in the first place.
Not exactly.
NT4 is Windows 4.0.
2000 is Windows 5.0
XP is 5.1
2003 is 5.2
XP SP2 and Server 2003 SP 1 are both 5.2
(using NTSwitch on XP gives you Whistler Server (2003 beta), on 2003 gives you XP SP1, on 2003 SP1 and XP SP2 switches between each other)
Among the Server and Workstation variants of each version number, all DLLs can be swapped about. The lesser variants are just missing DLLs and linked INFs in the installer that comes on the CD.
You can't swap DLLs between revisions, stuff breaks. NT4 cannot become 2003. But you can certainly convert Workstation to Server and back. All ntswitch does is change two registry entries... this is what makes a system behave as intended.
The missing files are just that... missing. You can drop them in and everything works fine. Nothing gets overwritten.
The exception is Datacenter/Cluster editions. The kernels on these editions are slightly different enough to make them unique. But there still are vastly many similarities.
That being said, you can see how excited I get when I get windows machines to do what I want. It's always an uphill battle. :-)
That's the most important reason why to install Server 2003 or XP. Once you start using it, it changes the way you work with Windows machines.
I suggest trying to find a copy of Server 2000 so at least you get Terminal Services (with unlimited connections in Per-User mode!). If you're too poor to spring for it, or don't trust P2P, you should try to find NTSwitch.exe... and follow these instructions:
Once you verify that Terminal Services is running and installed, you can revert the machine to Professional (or keep it at Server if you find it useful).
Seeing a 2K professional machine running multiple Terminal Services sessions without protest is a clear indication that the Server vs. Workstation distinction is only for market segmentation and maximizing profit, not any technical/support reason.
At least in my area, they've replaced the hardware on the street corner. We now have fiber coming into the junction box down the street, and copper going to each house. They've moved to using a cheaper, fewer-node headend a few hundred feet away, instead of having a large installation in the distribution facility. I imagine this headend works like a switch, not a hub, because I can't sniff anything.
And they've switched to randomly assigning DHCP addresses with non-overlapping subnets (very frustrating for in-house networking). Sigh.
This is coming from someone who already pays for cable TV.
Streaming multicast video from local networks? It'd be like having my own satellite feed. CNN Pipeline and other current video-on-demand stuff is a weak attempt; unambitious and ultimately flawed in execution.
See for example:0 .5/mtmail.html
http://www.zonageek.com/software/files/mt/mtmail-
Anyone can blog from anywhere.
There are RSS->blog gateways, and SMTP->RSS gateways.
At some point someone's going to get clever and collapse all these concepts into "message atoms". Descriptive text, along with tagged URLs and attachments that are treated as a unit with an author, publish date, keywords, "parent atom" for replies, etc.
Weblog, forum, RSS feed, email, XMPP (Jabber, Google Talk)... these are all just retrieval/display methods.
The future of blogging is when a standard gets created (similar to the SMTP MIME envelope standard or XMPP) that appropriately captures this concept and such that all such instances of it can be cast into the standard.
Then create gateways and display systems, database schemas, etc. that can handle these atoms and give us true independance from the medium and increased focus on the message.
The gconf configuration contains a lot of strings that applications use for help text and dialog boxes where options are displayed. Usually they are packaged with _every_ translation included. So the files can be huge if they have 100 copies of every string. Plus it's XML so it's wordy.
ogle and vlc support DVD nav, and you can jump to the root menu at any time (or start at any title).
Contrast this with, say, PowerDVD or some other Windows/MAC DVD player which makes you sit through the parts you don't want to.
Well, I would argue that magtape is a much better medium for capturing audio than a record, because the record has the disadvantage of being a physical process (needle being dragged around in resin) while the tape recording has little system reactivity (it doesn't take much EMF to line up domains on a thin film of tape).
That's why analog recording sessions are mastered to tape. Not vinyl.
Vinyl just so happened to be easy to reproduce, ship, and store. They suck as a reproductive medium. While the channel capacity is potential very high, the SNR is high as well, and the creation process introduces some non-tonal distortions that are difficult to filter or compensate for on playback.
Give me a digital reproductive medium anyday. Lasts forever and a good downmix speaks louder than words.
Here's an example:
http://www.un4seen.com/download.php?6chan
You can convert Dolby AC3 from DVDs into 5.1 channel OGG Vorbis streams to make DVD rips but it doesn't save you much bandwidth since AC3 is only 192kbps to begin with (can you believe it?)
I have a really nice OGG of the THX intro sound in 5.1 which sounds _much better_ than the AC3 version because of the higher bitrate ceiling. If I can find it I'll reply with a post to a link.
I don't have the slightest hesitation in using it to produce software that I might end up selling to someone else.
But I would never turn around and expect the same level of service that I would expect as a personal or hobbyist user.
If I was using it while working for a corporation I'd ask for the commercial license and also push for yearly maintenance, just to cover our collective asses.