Backwards compat is over-rated. It was a selling point for the PS2. I have a PS2, and also have a PS1. I have a library of over 40 genuine PS1 games. Know how often i've used the backwards compat in the past 9 years? About 3 times. And i have an old PS1 in the closet that I can use if i get desperate anyway (ditto for PS2 when i upgrade to a PS3). And even if i didn't i could pick up a PS2 for about 20-50 bucks these days anyway...
Except when you rename the PC you've destroyed any connection between the physical asset and any configuration database you have, such as a support history, purchasing, virus scanner database history, etc. Also, youv'e left an AD computer account that is no longer active in your directory that will need to be cleaned up (is the inactive computer account for that PC in storage, or has it been rebuilt??), and made it harder to keep track of volume licenses, etc.
Whatever naming scheme you choose, ensure that you can leave the names alone once they're assigned. Renaming PCs is bad and creates additional workload for no good reason.
It might not seem complicated, but there are a number of traps for new players. Most of these traps involve trying to store location/user/OS information in the hostname - which seems like a good idea at the time, but just gives you false information down the track when people quit, machines move, or the OS gets upgraded.
If you rename the PCs you're forever trying to keep up - or dealing with false information, which is worse than no information...
Exactly. You can get the location information out of AD (either the site, or domain name or IP address, or whatever). You can get the user by looking at who is logged in. You can get make/model information via WMI.
There's no spectacularly *good* workstation naming convention, but the service tag is just convenient - its stuck on the side of the box from the factory and retrievable from the bios come re-imaging time:)
Servers - sure, use a descriptive name for what it does or soem novelty name, so long as you're not planning on having a few hundred servers (which might sound a lot but many people do). Workstations are just too common for that though. Maybe on a small network you'll get away with it, but keep in mind that any successful company's small network will eventually become a big network, or at least end up with replacement hardware... eventually you'll either end up duplicating names, or run out.
Problem with that is that you will continually either have out of date PC names that are named according to where they AREN'T - or you need to continually rename PCs, thus completely ass-raping any configuration database you have (issue tracking, asset tracking, software licensing, virus scanner history, etc).
Renaming PCs = BAD. You get away with it up to a certain size, but once you start implementing apps like a job tracking system, software licensing tracking, etc it just bites you in the arse... HARD.
There is very little you can store in a workstation name that will be static and useful once you go beyond about 10 machines (maybe even less than that).
People move, machines get re-allocated, rebuilt, etc.
I use the service tag. Why? Several reasons:
its already printed on the machine
you can get it out of the bios when imaging the PC
its one less thing to ask the user for if you need to do a warranty claim
it will never change
if will be unique, presuming you are a single supplier organisation
Stuff like "bob-pc" or "accounts1" does not scale and either becomes inconsistent, or you need to keep renaming PCs which presents other issues (fucks up any configuration databases you have, etc).
So, service tag - boring as fuck, but does the job.
NDA. If your stuff is that important that a leak would be a really bad thing, ensure that you're able to be compensated appropriately for it.
Bear in mind that there's nothing to stop an angry local administrator stealing/selling data, and being more intimately involved with the company's business activities, he probably knows better where to look.
But, I'd suggest not outsourcing if posssible for a different reason. It normally doesn't work. The lack of local site knowledge is hugely detrimental to knowing wtf is going on. I was with a large aussie mining company that tried it - after 18 months they couldn't get away from the outsourcer fast enough. Main problems are that there is usually no continuity in who deals with a problem, no sense of personal responsibility, no problem ownership, and any admin who gets a clue at the outsourcer leaves and gets a real job as soon as they can.
You'll end up dealing with muppets who either don't care, have no clue, or both.
I guess the point of that is - why relicense it under the GPL? There was no need, as BSD code is permitted to be used anywhere be it in GPL products or commercial.
It was already BSD licensed, all relicensing under the GPL did was take away people's ability to use it in commercial products - which is NOT what the original author and copyright owner intended.
There's still an incentive to contribute to BSD licensed code.
What's that you ask? Its the ability to keep updated when the originator of the code updates his source. If you get your changes integrated into the main tree, you don't have to keep re-porting your additions to it, and dealing with the associated debugging, etc.
.. and some people want to actively encourage this to promote the spread of good standard libraries, etc.
How do you think things would have worked out of TCP/IP was GPLed? I'll tell you how - we would not have the inter-operability between many different pieces of network hardware that we have today...
Ahh you drank the RMS cool-aid. GPL is not free, as it takes away your ability to use the code in commercial products whilst keeping your enhancements closed-source.
Re:This isn't sensationalist, it's the truth
on
Leaving the GPL Behind
·
· Score: 4, Interesting
Exactly. There's a clear difference between GPL zealots and BSD zealots (of which, I am one).
GPL people want code to be publicly available no matter what.
BSD people just want the code to be used for the good of society; if you make money off it, so what.
If TCP/IP was originally GPLed, the protocol would probably not have taken off. BSD licensing (or public domain) promotes standards propogation. GPL encourages reinvention of the wheel, when someone decides that (heaven forbid) they want to be paid for their work (be it huge code additions, or just packaging up free code in a nicer package, or whatever).
Me? I'd much rather there was freely available BSD code for whatever problems have already been solved. Commercial products would be free to implement this well tested, standards compliant code and provide additional features that others may or may not be interested in. Those not interested in paying could take the base, well tested code and write their own pretty interface (or whatever) for it.
Someone subscribed to TechNet or MSDN is not going to be running Windows 7 on a Mac.
Why not?
I, for one, have a Mac and a technet subscription - the only reason I don't have the RTM downloading right now is because I've only just signed up and am waiting for my account to be processed.
ANYONE is free to get a technet subscription and get access to any and all microsoft products (for time-unlimited "evaluation" on unlimited machines for personal use) for about the same price as a copy of Windows ultimate.
If you're a geek (i.e., the type of person to run bootcamp) and want a copy of Windows RTM now, its "generally available" so long as you have broadband and about 300-400 bucks.
My experience with windows in the past 15 years or so does not concur. it sounds very much like you do not have DMA turned on, on your hard drive controller.
I actually noticed that disk i/o was better (as in, did affect the rest of the system as much) in NT4 than 95 or 98...
- There is nothing "non-standard" in QT and GTK. Every single linux distribution since 1994 will provide them.
Which version? Do GTK and QT provide APIs for database access, network connectivity, HTML rendering, etc? No... you need Gnome or KDE for that, and they're still in flux.
GTK has broken backwards compatibility just ONCE in 11 years, when switching from gtk 1.2 to gtk 2.0. Applications based on gtk 1 continue to run flawlessly in current distributions.
- QT has broken backwards compatibility 3 times since 1991. The last time it has done so, during the 3.3 -> 4.0 switch, it provided a compile-time source compatibility layer. I no longer have 2.x apps around, but I can assure that 3.x applications keep running with no problem.
See above. I'm talking KDE and Gnome not QT and GTK. GTK and QT do not provide all the frameworks required for application development using standard libraries for stuff other than UI widgets. Gnome and KDE are nearer the mark, and they are continually breaking shit and changing (for better or worse).
There is no equivalent to appkit (for example) that you can RELY ON to be installed on a linux box as yet. There is no single platform to target. GTK and QT simply do not compare as an "application development framework", they're little more than widget libraries for GUI development. There is a lot more to applicate development than GUI widgets...
Linux has no stable, standard GUI application development platform for a start. Yes, KDE and gnome both exist, but no they're not "standard" and the API is still changing and breaking backwards compatibility regularly.
Go play with Cocoa / Xcode / Interface builder, and you'll get a bit of an idea as to why Linux is even now, still trying to catch up to NextSTEP 1991.
This is why there is a lack of high quality applications.
Don't get me wrong, Linux is great and I'm trying to get into OpenStep development myself (so i can do OS/X -> Free unix cross platform application development), but the state and lack of standardization on toolkits on Linux is quite apparent.
Libraries and frameworks. You know, bits of software that programmers can use to ensure a consistent feel and program behavior. OS/X has them. Windows has them. Linux is still working on it (KDE/Gnome are getting there, but shit still breaks in a major way from version to version).
The stuff you can't see with the benchmarks, that people actually notice and care about in reality. Like UI responsiveness. Seriously, the RC is still available, go download it and check it out rather than speculating wildly.
Slashdot Mirror
For values of "upgrade" that mean wipe, reinstall, user state migrate, yes.
Backwards compat is over-rated. It was a selling point for the PS2. I have a PS2, and also have a PS1. I have a library of over 40 genuine PS1 games. Know how often i've used the backwards compat in the past 9 years? About 3 times. And i have an old PS1 in the closet that I can use if i get desperate anyway (ditto for PS2 when i upgrade to a PS3). And even if i didn't i could pick up a PS2 for about 20-50 bucks these days anyway...
Whatever naming scheme you choose, ensure that you can leave the names alone once they're assigned. Renaming PCs is bad and creates additional workload for no good reason.
If you rename the PCs you're forever trying to keep up - or dealing with false information, which is worse than no information...
There's no spectacularly *good* workstation naming convention, but the service tag is just convenient - its stuck on the side of the box from the factory and retrievable from the bios come re-imaging time :)
Servers - sure, use a descriptive name for what it does or soem novelty name, so long as you're not planning on having a few hundred servers (which might sound a lot but many people do). Workstations are just too common for that though. Maybe on a small network you'll get away with it, but keep in mind that any successful company's small network will eventually become a big network, or at least end up with replacement hardware... eventually you'll either end up duplicating names, or run out.
Renaming PCs = BAD. You get away with it up to a certain size, but once you start implementing apps like a job tracking system, software licensing tracking, etc it just bites you in the arse... HARD.
People move, machines get re-allocated, rebuilt, etc.
I use the service tag. Why? Several reasons:
Stuff like "bob-pc" or "accounts1" does not scale and either becomes inconsistent, or you need to keep renaming PCs which presents other issues (fucks up any configuration databases you have, etc).
So, service tag - boring as fuck, but does the job.
Bear in mind that there's nothing to stop an angry local administrator stealing/selling data, and being more intimately involved with the company's business activities, he probably knows better where to look.
But, I'd suggest not outsourcing if posssible for a different reason. It normally doesn't work. The lack of local site knowledge is hugely detrimental to knowing wtf is going on. I was with a large aussie mining company that tried it - after 18 months they couldn't get away from the outsourcer fast enough. Main problems are that there is usually no continuity in who deals with a problem, no sense of personal responsibility, no problem ownership, and any admin who gets a clue at the outsourcer leaves and gets a real job as soon as they can.
You'll end up dealing with muppets who either don't care, have no clue, or both.
It was already BSD licensed, all relicensing under the GPL did was take away people's ability to use it in commercial products - which is NOT what the original author and copyright owner intended.
What's that you ask? Its the ability to keep updated when the originator of the code updates his source. If you get your changes integrated into the main tree, you don't have to keep re-porting your additions to it, and dealing with the associated debugging, etc.
How do you think things would have worked out of TCP/IP was GPLed? I'll tell you how - we would not have the inter-operability between many different pieces of network hardware that we have today...
Ahh you drank the RMS cool-aid. GPL is not free, as it takes away your ability to use the code in commercial products whilst keeping your enhancements closed-source.
If TCP/IP was originally GPLed, the protocol would probably not have taken off. BSD licensing (or public domain) promotes standards propogation. GPL encourages reinvention of the wheel, when someone decides that (heaven forbid) they want to be paid for their work (be it huge code additions, or just packaging up free code in a nicer package, or whatever).
Me? I'd much rather there was freely available BSD code for whatever problems have already been solved. Commercial products would be free to implement this well tested, standards compliant code and provide additional features that others may or may not be interested in. Those not interested in paying could take the base, well tested code and write their own pretty interface (or whatever) for it.
Why not?
I, for one, have a Mac and a technet subscription - the only reason I don't have the RTM downloading right now is because I've only just signed up and am waiting for my account to be processed.
ANYONE is free to get a technet subscription and get access to any and all microsoft products (for time-unlimited "evaluation" on unlimited machines for personal use) for about the same price as a copy of Windows ultimate.
If you're a geek (i.e., the type of person to run bootcamp) and want a copy of Windows RTM now, its "generally available" so long as you have broadband and about 300-400 bucks.
I actually noticed that disk i/o was better (as in, did affect the rest of the system as much) in NT4 than 95 or 98...
Windows 7 did include a scheduler re-write. They got rid of the dispatcher lock (whatever that is). All i know is its a lot smoother.
See sig. Clown.
"application" development, obviously... *sigh*
Which version? Do GTK and QT provide APIs for database access, network connectivity, HTML rendering, etc? No... you need Gnome or KDE for that, and they're still in flux.
See above. I'm talking KDE and Gnome not QT and GTK. GTK and QT do not provide all the frameworks required for application development using standard libraries for stuff other than UI widgets. Gnome and KDE are nearer the mark, and they are continually breaking shit and changing (for better or worse).
There is no equivalent to appkit (for example) that you can RELY ON to be installed on a linux box as yet. There is no single platform to target. GTK and QT simply do not compare as an "application development framework", they're little more than widget libraries for GUI development. There is a lot more to applicate development than GUI widgets...
I've been running 7 for the past 4 months, mate. Have you?
Go play with Cocoa / Xcode / Interface builder, and you'll get a bit of an idea as to why Linux is even now, still trying to catch up to NextSTEP 1991.
This is why there is a lack of high quality applications.
Don't get me wrong, Linux is great and I'm trying to get into OpenStep development myself (so i can do OS/X -> Free unix cross platform application development), but the state and lack of standardization on toolkits on Linux is quite apparent.
Libraries and frameworks. You know, bits of software that programmers can use to ensure a consistent feel and program behavior. OS/X has them. Windows has them. Linux is still working on it (KDE/Gnome are getting there, but shit still breaks in a major way from version to version).
It will run fine with $50 worth of RAM (ie, 2gb or more).
The stuff you can't see with the benchmarks, that people actually notice and care about in reality. Like UI responsiveness. Seriously, the RC is still available, go download it and check it out rather than speculating wildly.