I haven't tried this in a while, but if you have access to the regedit.exe program, then you can get a text dump of the registry. It's a pretty darn huge file (65 MB on my old Win95 station).
The MS approved method for defragging the registry, in fact, was to boot to DOS, get a text dump of the registry, blow away the user.dat and.dat files that made up the binary registry, then import the text dump.
So, if you want to know everything that is in the registry, just get a text dump of it, compress it down to virtually nothing, and cart it away on diskette for later reading.:)
Actually, running this configuration can give plenty of benefits. Yes, you still have to purchase the Windows software, but you don't have to purchase skookum workstations to run them on, nor do you have to purchase expensive MS server software to run a similar setup.
Buy one skookum server for $3000 or so (dual-proc, mega ram, good disks). Then run thin-clients and diskless workstations. Depending on the speed of the client CPU and the amount of RAM on the client, you can either run pure-remote (run app on server, show display on client) or local-mode (download app to client ram, run on client cpu), and even run a combination of the two on the same client.
You have super-stable workstations running the normal Windows software. You only have to administer a single software image (the server) and upgrades are simple as there is only a single server image to worry about. All the clients boot remotely across the network, and the local harddrive is used solely for swapspace.
It's a great concept, and an IT admin's dream -- only 1 box to worry about, full control over everything. All you have to worry about now is dying hardware on the clients.:)
Depending on your clients (this won't work with 486s or Pentiums) you can use "Local Mode" or even mixed Local/Remote. In this setup, the client downloads the app to RAM, then runs it from there using the local CPU, local RAM, local vid card, local sound card, and so on. Relieves the load from the server, but still allows you to run diskless clients (although we use the local HD for swap space).
We're using this setup in our elementary school Linux labs. We are running CodeWeaver's version of WINE for the odd Windows app that does not yet have a Linux counterpart. We haven't told them that it's the Windows version running, though, as then they'd want to run other Windows apps and we'd be no better off than before the switch to Linux.:)
This is one of the major issues we are facing with implementing Unix/OSS in high schools. All the teachers are of the mindset "We have to teach MS Office, it's what's used in the business world." And that really irritates me.
Do we teach people how to use a single piece of software (ie. Word), or do we teach people concepts, skills, and abilities so that they can learn to use any piece of software in a very short period of time? Do we teach them where MS decides the "Create Table" menu item should be, or do we teach them how to make nice looking tables and charts, regardless of what software they use?
We should be doing the latter. We should be teaching people to teach themselves, to become learners, self-sufficient. We should not be teaching people how to memorise useless crud like where the "Bold" button is in Word, nor should we be teaching them how to use a single application. If they want to learn a specific application in-depth, then they should take courses designed specifically for that. School is about learning how to learn, not about learning how app X works.
Times are changing, however. There are several school district in BC, Canada that are moving toward an X Terminal setup in school computer labs. For instance, SD73 has implemented 37 Linux labs in elementary schools, and is piloting a CAD lab in a secondary.
There are also school district in Oregan which are moving Linux into secondary school labs.
Sure, these aren't super-huge districts in super-huge cities, but the movement has begun.
Now, if only we could convince them all to drop Linux in favour of FreeBSD. Then we'd see the administration costs drop even further as sysadmins no longer have to fight dependency-hell, illogical filesystem, incompatibilities amongst distros, and all the other "niceties" that come with Linux.:)
This setup makes administration of software/network related issues a breeze. This is the model School District 73 in Kamloops, BC, Canada is using. 37 elementary school labs running over 1500 desktops, all running Linux. Most of the hardware was donated from the governement or purchased used via Computers for Schools.
Remote administration via SSH and VNC/remote X is a snap, and most problems are solved interactively with the teacher in the school, reducing downtime. Updates are a matter of scripted SSH sessions afterhours.
The only downside is that it is all done using Linux. But, that's a personal bias, as I find Linux to be horribly disorganised, disfunctional, and a royal pain to work with compared to FreeBSD. That's a discussion for another time, though.:)
The students all love the system, the administration loves the system, the software techs like the system more and more as they use it, and the teachers love it as well. And it does save money in the long run: $10-20,000 in Novell licenses, $10-20,000 in MS Licenses, per year. And the time savings are enormous.
Except the MS filesystem will be a gian SQL database running on a slimmed-down SQL Server, meaning the hardware requirements for Longhorn will be through the roof.
Uh, you can use *ANY* IMAP server with SquirrelMail. In fact, it even comes with presets for Courier, UW-IMAP, Cyrus, and Exchange. It can also be used with *ANY* MTA, so long as the MTA and the IMAP server both use the same storage methods (you can't use mbox format with Courier-IMAP, for instance, so you can't use Sendmail with Courier-IMAP).
I've got several installations of SM using Sendmail + UW-IMAP, and several with Postfix + Courier-IMAP. Works great. I've used a few that used Sendmail + Cyrus as well.
Actually, compiling XFree86, KDE, or the entire FreeBSD OS doesn't take all that long on a 333MHz box. It all comes down to time management: set it to do everything overnight (or overday) while you sleep.:)
Where having more CPU power comes into play is with real-time anything, image/audio/video editing, server apps, and games playing. Everything else runs fine on a P-166 MHz box.
The best thing you can do to better your helpdesk is to hire people who actually know what they are dealing with. Don't hire punks straight out of high school and give them a checklist to walk through for each call... nobody will use the helpdesk then.
Hire young punks still in Uni, train them in the use of the software they are supporting, sit them down with the developers to see what the helpdesk can handle and what the developers should handle, sit them down with any other tech support people to figure out what each group should handle, and give them a couple of boxes to bang around on with the software they are looking after.
Don't treat your helpdesk people like morons, and the users won't see them that way.
The problem with IE on windows was that you could not remove IE, nor could you upgrade it without upgrading your Windows system.
Having an HTML rendering engine as part of the OS can make sense if done right. Use it for displaying help pages, use it for displaying HTML messages, make it available for third-party browsers to use for web browsing. But don't make the main browser a part of the OS such that upgrading it upgrades your system.
There is *NO* difference between Windows Explorer the file manager, Explorer the shell, and Internet Explorer the web browser. They are the same binary, and the libs are system DLLs that cannot be removed.
If you don't like Safari, just delete it. If you don't like Opera, just delete it. If you don't like Mail.app, just delete it. If you don't like IE on Windows, too bad, you're stuck with it, and all it's bugs and security issues.
Actually, the prime difference is that MS made IE so intertwined with Windows that upgrading one is the same as upgrading the other. You cannot upgrade IE without also upgrading a lot of Windows components and system dlls.
Upgrading Safari will be as simple as installing a new binary. Don't like the new version, delete it and install the old version. Whoop-de-do. There's no behind-the-scenes upgrades to the OS or anything like that.
God, I hope that a "good browser" is *not* one that displays broken websites. A "good browser" is one that displays an error message when it comes across non-standard HTML, CSS, XML, etc. It should show a nice error message stating that the site is non-standard, that it has no way of knowing what the webmaster was trying to do, and that it cannot display anything.
The reason we have bloated, incompatible browsers is that there is no method of checking for validated, conforming, well-formed HTML pages. Once XHTML starts being used more, we'll see a lot of these error messages start to popup, and the web will become a much better place.
FreeBSD is great, if you are willing to take the time to learn. The man pages are up-to-date and useful, so don't be afraid to use them (you'll definitely want to read the pages in section 7). There are lots of online resources on the FreeBSD website, don't be afraid to use them.
Learn to use the ports tree, and learn to really use it. Read the Porter's Handbook to really understand how it works, it'll save you a lot of time down the road, especially when doing updates.
FreeBSD makes a great system, whether it is a desktop running X Window, or a server running a bunch of services. (I've got 5.0-RC2 running on my laptop, and all I had to tweak was the XF86Config to use the right savage driver -- even hibernate works.)
You need the ZAxisMapping option as well in order to get the scroll wheel/button working.
And for those that don't want a mouse at the console, don't start moused, and change the mouse device in XF86Config to point to the mouse device (/dev/psm0,/dev/ums0,/dev/sio0, and so on).
In other words, tabs work completely different from the standard MDI.
I'm not missing the point. I'm trying to show that tabs are not the panacea, nor the great missing link that everyone keeps trying to make them out to be.
There's nothing wrong with a browser that does not include tabs. IMO, there's plenty of things wrong with a browser that includes tabs a la Mozilla, as it is a very poor implementation. More Mozilla users need to use Opera to see just how bad Mozilla does things.
Can you eliminate the tab bar? Can you minimise/maximise individual tabs? Can you resize tabs? Can you put more than one tab onscreen at a time (tile/cascade)? Can you use standard keys for moving between tabs (CTRL+TAB, CTRL+SHIFT+TAB)? Can you use standard keys for creating (CTRL+N) and closing (CTRL+F4) tabs?
Last time I tried Mozilla (1.0 or 1.0.1) and Netscape (6.02), you couldn't. All you could do is open or close tabs, using a completely separate set of keys for manipulating tabs, and everything still revolved around the "one window per page" SDI.
Maybe things have changed since the days of 1.0. Hopefully they have. But, I've given up on Mozilla for the foreseeable future.
What's so great about tabs? Opera has had them for aeons, and I've found them to be nothing but trouble. They take up screen real estate. The require you to use the mouse. They are a pain to close (Mozilla has a horrible interface). And they don't do anything particularly useful that a good MDI (or well organised SDI) interface doesn't already provide.
Same with the Sidebar. I've yet to find a single, useful implementation of a sidebar that doesn't use up too much space for too little value (Nautilus is the worst for this, but Netscape, and Mozilla aren't pretty crummy as well).
What would be better is if more apps in general followed Opera's lead and made better use of the MDI. One window for the app, separate windows for each "page" within that window. Key bindings for everything so that you don't need to use the mouse except when absolutely necessary. But also use mouse gestures so that those who are dependent (poor sods) on the mouse never need to use the keyboard.
Opening separate application windows for each "page" is an absolute waste and one reason I refuse to use most MS products now. Tabs are a poor man's MDI and all developers who find them "new, innovative, useful" should be shot.
You want Knoppix. It's a bootable CD-ROM that takes you right into KDE. So long as you have a fairly recent PC with decent components, all you do is pop in the CD and start using it. If you need to save custom config changes, just save them to a floppy.
It's not so much that you show every single step in the process, but that you show you understand the mechanics behind the problems. Whether you do the math in your head, on paper, or whatnot shouldn't have any bearing on the marks granted.
I usually just wrote down the answer, and sometimes the odd formula that I used, but never wrote down all the boring number crunching I did, as that would take too long. It was all done in my head, anyway. By the time I had written out the formulas, I had the answer.
This also annoyed my science profs during unit conversion exercises. I'd just write out the bare equations and conversions, and do all the number math in my head.
I hated teachers like this. And teachers like this didn't like me much. I did most of my work in my head while writing the problem down. By the time I had finished writing out the original equation, I had the answer. On those questions where I needed to take more than a single step, I usually combined multiple steps. This really irritated the teachers.
On those tests where I forced myself to write out the steps, I would routinely combine multiple steps into one, as that is the way my mind works. You could see the steps I had taken, but, since the answer was supposed to be written out as a 7-step / 7-line answer, and mine was only 3 lines long, I lost marks. Nevermind that each step is shown in the answer, or that the answer is correct, or that the logic used to get the answer is correct.
I was also quite good at coming up with shortcuts and new formulas that would save time on certain classes of problems. I lost a lot of marks and had many an argument with teachers over this, as well. Just because I didn't use the exact same logic as the teacher is no reason for me to be punished.
The problem with this kind of system (show your work, you'll get partial marks) is that it teaches students to become bullshitters.
For a business that may already have some 802.11b gear, it'd be best to wait for 802.11g. It uses most of the algorithms from 802.11a for bandwidth and encryption, but it is compatible with 802.11b, making their existing investment in wireless technology useful.
802.11a also doesn't have the greatest range, especially indoors. 802.11b and 802.11g have much better range (and 802.11g has the same throughput as 802.11a).
Slashdot Mirror
I haven't tried this in a while, but if you have access to the regedit.exe program, then you can get a text dump of the registry. It's a pretty darn huge file (65 MB on my old Win95 station).
.dat files that made up the binary registry, then import the text dump.
:)
The MS approved method for defragging the registry, in fact, was to boot to DOS, get a text dump of the registry, blow away the user.dat and
So, if you want to know everything that is in the registry, just get a text dump of it, compress it down to virtually nothing, and cart it away on diskette for later reading.
Actually, running this configuration can give plenty of benefits. Yes, you still have to purchase the Windows software, but you don't have to purchase skookum workstations to run them on, nor do you have to purchase expensive MS server software to run a similar setup.
:)
Buy one skookum server for $3000 or so (dual-proc, mega ram, good disks). Then run thin-clients and diskless workstations. Depending on the speed of the client CPU and the amount of RAM on the client, you can either run pure-remote (run app on server, show display on client) or local-mode (download app to client ram, run on client cpu), and even run a combination of the two on the same client.
You have super-stable workstations running the normal Windows software. You only have to administer a single software image (the server) and upgrades are simple as there is only a single server image to worry about. All the clients boot remotely across the network, and the local harddrive is used solely for swapspace.
It's a great concept, and an IT admin's dream -- only 1 box to worry about, full control over everything. All you have to worry about now is dying hardware on the clients.
Depending on your clients (this won't work with 486s or Pentiums) you can use "Local Mode" or even mixed Local/Remote. In this setup, the client downloads the app to RAM, then runs it from there using the local CPU, local RAM, local vid card, local sound card, and so on. Relieves the load from the server, but still allows you to run diskless clients (although we use the local HD for swap space).
:)
We're using this setup in our elementary school Linux labs. We are running CodeWeaver's version of WINE for the odd Windows app that does not yet have a Linux counterpart. We haven't told them that it's the Windows version running, though, as then they'd want to run other Windows apps and we'd be no better off than before the switch to Linux.
This is one of the major issues we are facing with implementing Unix/OSS in high schools. All the teachers are of the mindset "We have to teach MS Office, it's what's used in the business world." And that really irritates me.
Do we teach people how to use a single piece of software (ie. Word), or do we teach people concepts, skills, and abilities so that they can learn to use any piece of software in a very short period of time? Do we teach them where MS decides the "Create Table" menu item should be, or do we teach them how to make nice looking tables and charts, regardless of what software they use?
We should be doing the latter. We should be teaching people to teach themselves, to become learners, self-sufficient. We should not be teaching people how to memorise useless crud like where the "Bold" button is in Word, nor should we be teaching them how to use a single application. If they want to learn a specific application in-depth, then they should take courses designed specifically for that. School is about learning how to learn, not about learning how app X works.
Times are changing, however. There are several school district in BC, Canada that are moving toward an X Terminal setup in school computer labs. For instance, SD73 has implemented 37 Linux labs in elementary schools, and is piloting a CAD lab in a secondary.
:)
There are also school district in Oregan which are moving Linux into secondary school labs.
Sure, these aren't super-huge districts in super-huge cities, but the movement has begun.
Now, if only we could convince them all to drop Linux in favour of FreeBSD. Then we'd see the administration costs drop even further as sysadmins no longer have to fight dependency-hell, illogical filesystem, incompatibilities amongst distros, and all the other "niceties" that come with Linux.
This setup makes administration of software/network related issues a breeze. This is the model School District 73 in Kamloops, BC, Canada is using. 37 elementary school labs running over 1500 desktops, all running Linux. Most of the hardware was donated from the governement or purchased used via Computers for Schools.
:)
Remote administration via SSH and VNC/remote X is a snap, and most problems are solved interactively with the teacher in the school, reducing downtime. Updates are a matter of scripted SSH sessions afterhours.
The only downside is that it is all done using Linux. But, that's a personal bias, as I find Linux to be horribly disorganised, disfunctional, and a royal pain to work with compared to FreeBSD. That's a discussion for another time, though.
The students all love the system, the administration loves the system, the software techs like the system more and more as they use it, and the teachers love it as well. And it does save money in the long run: $10-20,000 in Novell licenses, $10-20,000 in MS Licenses, per year. And the time savings are enormous.
Except the MS filesystem will be a gian SQL database running on a slimmed-down SQL Server, meaning the hardware requirements for Longhorn will be through the roof.
Uh, you can use *ANY* IMAP server with SquirrelMail. In fact, it even comes with presets for Courier, UW-IMAP, Cyrus, and Exchange. It can also be used with *ANY* MTA, so long as the MTA and the IMAP server both use the same storage methods (you can't use mbox format with Courier-IMAP, for instance, so you can't use Sendmail with Courier-IMAP).
I've got several installations of SM using Sendmail + UW-IMAP, and several with Postfix + Courier-IMAP. Works great. I've used a few that used Sendmail + Cyrus as well.
.
Actually, compiling XFree86, KDE, or the entire FreeBSD OS doesn't take all that long on a 333MHz box. It all comes down to time management: set it to do everything overnight (or overday) while you sleep. :)
Where having more CPU power comes into play is with real-time anything, image/audio/video editing, server apps, and games playing. Everything else runs fine on a P-166 MHz box.
The best thing you can do to better your helpdesk is to hire people who actually know what they are dealing with. Don't hire punks straight out of high school and give them a checklist to walk through for each call ... nobody will use the helpdesk then.
Hire young punks still in Uni, train them in the use of the software they are supporting, sit them down with the developers to see what the helpdesk can handle and what the developers should handle, sit them down with any other tech support people to figure out what each group should handle, and give them a couple of boxes to bang around on with the software they are looking after.
Don't treat your helpdesk people like morons, and the users won't see them that way.
It's really quite simple.
The problem with IE on windows was that you could not remove IE, nor could you upgrade it without upgrading your Windows system.
Having an HTML rendering engine as part of the OS can make sense if done right. Use it for displaying help pages, use it for displaying HTML messages, make it available for third-party browsers to use for web browsing. But don't make the main browser a part of the OS such that upgrading it upgrades your system.
There is *NO* difference between Windows Explorer the file manager, Explorer the shell, and Internet Explorer the web browser. They are the same binary, and the libs are system DLLs that cannot be removed.
If you don't like Safari, just delete it. If you don't like Opera, just delete it. If you don't like Mail.app, just delete it. If you don't like IE on Windows, too bad, you're stuck with it, and all it's bugs and security issues.
Actually, the prime difference is that MS made IE so intertwined with Windows that upgrading one is the same as upgrading the other. You cannot upgrade IE without also upgrading a lot of Windows components and system dlls.
Upgrading Safari will be as simple as installing a new binary. Don't like the new version, delete it and install the old version. Whoop-de-do. There's no behind-the-scenes upgrades to the OS or anything like that.
God, I hope that a "good browser" is *not* one that displays broken websites. A "good browser" is one that displays an error message when it comes across non-standard HTML, CSS, XML, etc. It should show a nice error message stating that the site is non-standard, that it has no way of knowing what the webmaster was trying to do, and that it cannot display anything.
The reason we have bloated, incompatible browsers is that there is no method of checking for validated, conforming, well-formed HTML pages. Once XHTML starts being used more, we'll see a lot of these error messages start to popup, and the web will become a much better place.
32-bit, 33MHz PCI has a total bandwidth of 133 megabytes per second.
USB 2.0 has a total bandwidth of 480 megaBITS per second, which is only 60 megaBYTES per second, not even half of the PCI bus.
FreeBSD is great, if you are willing to take the time to learn. The man pages are up-to-date and useful, so don't be afraid to use them (you'll definitely want to read the pages in section 7). There are lots of online resources on the FreeBSD website, don't be afraid to use them.
:)
Learn to use the ports tree, and learn to really use it. Read the Porter's Handbook to really understand how it works, it'll save you a lot of time down the road, especially when doing updates.
FreeBSD makes a great system, whether it is a desktop running X Window, or a server running a bunch of services. (I've got 5.0-RC2 running on my laptop, and all I had to tweak was the XF86Config to use the right savage driver -- even hibernate works.)
Jump in, RTFM, and have fun.
You need the ZAxisMapping option as well in order to get the scroll wheel/button working.
/dev/ums0, /dev/sio0, and so on).
And for those that don't want a mouse at the console, don't start moused, and change the mouse device in XF86Config to point to the mouse device (/dev/psm0,
In other words, tabs work completely different from the standard MDI.
I'm not missing the point. I'm trying to show that tabs are not the panacea, nor the great missing link that everyone keeps trying to make them out to be.
There's nothing wrong with a browser that does not include tabs. IMO, there's plenty of things wrong with a browser that includes tabs a la Mozilla, as it is a very poor implementation. More Mozilla users need to use Opera to see just how bad Mozilla does things.
Can you eliminate the tab bar? Can you minimise/maximise individual tabs? Can you resize tabs? Can you put more than one tab onscreen at a time (tile/cascade)? Can you use standard keys for moving between tabs (CTRL+TAB, CTRL+SHIFT+TAB)? Can you use standard keys for creating (CTRL+N) and closing (CTRL+F4) tabs?
Last time I tried Mozilla (1.0 or 1.0.1) and Netscape (6.02), you couldn't. All you could do is open or close tabs, using a completely separate set of keys for manipulating tabs, and everything still revolved around the "one window per page" SDI.
Maybe things have changed since the days of 1.0. Hopefully they have. But, I've given up on Mozilla for the foreseeable future.
What's so great about tabs? Opera has had them for aeons, and I've found them to be nothing but trouble. They take up screen real estate. The require you to use the mouse. They are a pain to close (Mozilla has a horrible interface). And they don't do anything particularly useful that a good MDI (or well organised SDI) interface doesn't already provide.
Same with the Sidebar. I've yet to find a single, useful implementation of a sidebar that doesn't use up too much space for too little value (Nautilus is the worst for this, but Netscape, and Mozilla aren't pretty crummy as well).
What would be better is if more apps in general followed Opera's lead and made better use of the MDI. One window for the app, separate windows for each "page" within that window. Key bindings for everything so that you don't need to use the mouse except when absolutely necessary. But also use mouse gestures so that those who are dependent (poor sods) on the mouse never need to use the keyboard.
Opening separate application windows for each "page" is an absolute waste and one reason I refuse to use most MS products now. Tabs are a poor man's MDI and all developers who find them "new, innovative, useful" should be shot.
802.11g is also backwards compatible with 802.11b gear.
It's a hybrid of 802.11b and 802.11a. It uses a lot of the same algorithms as 802.11a, but in the same 2.4 GHz spectrum that 802.11b uses.
Works fine with IE 6 on Windows 98SE. Guess it just doesn't like you. :)
You want Knoppix. It's a bootable CD-ROM that takes you right into KDE. So long as you have a fairly recent PC with decent components, all you do is pop in the CD and start using it. If you need to save custom config changes, just save them to a floppy.
Here, here. I agree.
It's not so much that you show every single step in the process, but that you show you understand the mechanics behind the problems. Whether you do the math in your head, on paper, or whatnot shouldn't have any bearing on the marks granted.
I usually just wrote down the answer, and sometimes the odd formula that I used, but never wrote down all the boring number crunching I did, as that would take too long. It was all done in my head, anyway. By the time I had written out the formulas, I had the answer.
This also annoyed my science profs during unit conversion exercises. I'd just write out the bare equations and conversions, and do all the number math in my head.
I hated teachers like this. And teachers like this didn't like me much. I did most of my work in my head while writing the problem down. By the time I had finished writing out the original equation, I had the answer. On those questions where I needed to take more than a single step, I usually combined multiple steps. This really irritated the teachers.
On those tests where I forced myself to write out the steps, I would routinely combine multiple steps into one, as that is the way my mind works. You could see the steps I had taken, but, since the answer was supposed to be written out as a 7-step / 7-line answer, and mine was only 3 lines long, I lost marks. Nevermind that each step is shown in the answer, or that the answer is correct, or that the logic used to get the answer is correct.
I was also quite good at coming up with shortcuts and new formulas that would save time on certain classes of problems. I lost a lot of marks and had many an argument with teachers over this, as well. Just because I didn't use the exact same logic as the teacher is no reason for me to be punished.
The problem with this kind of system (show your work, you'll get partial marks) is that it teaches students to become bullshitters.
For a business that may already have some 802.11b gear, it'd be best to wait for 802.11g. It uses most of the algorithms from 802.11a for bandwidth and encryption, but it is compatible with 802.11b, making their existing investment in wireless technology useful.
802.11a also doesn't have the greatest range, especially indoors. 802.11b and 802.11g have much better range (and 802.11g has the same throughput as 802.11a).