In is not just the new drivers (teens). A large number of drivers are really unsafe. In North America we treat driving as a right instead of the privilege it is.
There is a big difference in having a 650 MB CDR sitting in your desk and 100 MB of storage space sitting up on an Apple server.
You are correct. The CDR is larger and more reliable.
If you had a fire? Theft? Water damage? All your computers (and disks) go with it. If Apple is hosting your back up, you can get it back.
This is silly. For a couple bucks a year I can cram a lot of offsite CDR backups in my safety deposit box. Failing that I can swap offsite storage with my buddy.
Any software, Sun's or MS's, gets insanely expensive in a hurry. If you took a 1000 person company and compared the true Total Cost of Ownership (TCO) of a complete Solaris system vs. a complete Win system you would probably find that the cost is the same. A company of this size should have support contracts for the sofware and contracts for hardware, plus a small army of admins. After those yearly costs the initial hardware and licence fees are peanuts. Even if the initial licence fee is $0 for a free Unix. Both Sun and MS have tried to prove that the TCO is in their favour. Neither have been particularly convinceing. You can easily replace "Sun" with any Unix vendor.
All vendors use per seat licenseing to some degree or another. You might be paying for a per seat license directly ala MS, or you might pay in upfront costs ala Sun. The per seat licenseing really only affects the little guys. Most of the bigger sites will just buy a site wide license if everyone is going to use the software. It should be noted that most of the interesting software is licensed per seat. All Sun really ships that is not per seat licenced is a fileserver and print spooler. If you use a free Unix you get a "complete exterprise solution" for free provided you have a small army of admin staff. This is were TCO kicks in. Licenseing is only a small part of TCO. It is the support contracts and admin staff that will really determine the TCO.
For any system that combines a mix of platforms the TCO will usually be higher. This is all this MS page really shows. A windows desktop is assumed.
Today due in large part to StarOffice/OpenOffice, Mozilla, and Evolution any office of any size could drop MS from the desktop and be quite happy. However the cost of switching remains really really high.
I agree. The whole intimidation thing is real kicker isen't it. The people I hate are the ones who want to be intimidated. They feel superior for being stupid. How does that work? Compare this to my grandfather who seems to do alright despite being half blind[1]. Due to his handicap I think he looks at anything new as a challenge, and plesantly surprised when it isen't. Other people are probably pissed that a computer is supposed to be easy to use, as per marketing, but isen't. It makes them feel stupid. Those of us with more experience aren't intimidated, and when a computer dosen't work we think the developers are stupid. Maybe this is what we should tell users?
Instead of trying to convince users that computers are easy to use do the opposite. Tell them that computers are a pain in the ass (the truth). Tell users that the user interface was "designed" by a bunch stupid punk assed condesending arrogant new hires whacked out on caffine (possibly true). Now when the user manages to remember the basics they will fell a sense of superiority and acomplishment. I am only half kidding.
Regardless of the true complexity I often start off a mini training session with "I don't know why they had to make this so complicated..." It works for some people. Others are really stupid.
[1] Setting up a computer for my grandfather is what started me thinking about how poor computer user interfaces really are. My grandfather has lost much of his vision, but can still read if the text is magnified to atleast 3cm/inch high letters. A typewrite has a small font. There is not much you can do about it. In theory a computer should be able to display any sized text even if there is only one letter per screen. Too bad theory dosen't work out. For many reasons it is much easier for him to use a simple magnifying lens then to try to fight with the OS and display big fonts. How peverted is that?
Part of the problem there is that the users have probably had bad experiences with manuals in the past. Most computer manuals I have tried to read are very poorly written. I can't blame a user for looking at a computer manual in disgust, as I do the same. In general, documentation abuses the language, is poorly organized, and incomplete.
The computer industry does not give the users any credit and assumes all users are idiots. This mindset has UI designers convinved that they have to map computers onto something familar (desktops). This makes as much sense as putting a stick shift on a toaster to control the darkness. It will work, but is it really the best interface? By attempting to map "familiarity" the designers only frustrate the users when the familiarity inevetibly breaks due to a bug or a poor metaphor. Nothing frustrates a user more then when the actual outcome is different from the expected outcome. You can't map a car's interface onto a computer. Nor can you compared a car's "easy" interface to a computer's. A computer is a completely different device.
Here lies the problem: instead of treating computers as the powerfull devices they are, designers instead attempt to "perfect" the poor metaphors adopted years ago. Computer interfaces could be far more powerfull then they are. Instead a computers spends most of its power by drawing pretty pictures. Computer interfaces should expose the user to the power in the computer. This dosen't mean tab upon tab of check box options. Designers have to find new ways to present data. Not pretty ways, but usefull ways.
Not that this will happen. It is much easier to sell an old program with a new look then it is to discover better ways to manipulate data.
There is a chance that your passphrase could end up in the swap file. Use an encrypted swap to be safe.
Re:Linux for desktop, *BSD for servers?
on
FreeBSD 4.6
·
· Score: 2
IIRC FreeBSD did this as well, during the install
True enough. I just did a FreeBSD install and it was a little different then I remember. I guess most of what I have written only applies to OpenBSD now.
Also what do you mean by top heavy concerning the config files?
One of the beauties of OpenBSD is the/etc/rc.conf file. It is a file of sh parameters that are sourced by the rc scripts at boot time. In it you have config options for most of the major services that you would run. For example if I want sendmail to run with "-q30m -bd" (run queue every 30mins, listen on port 25) I set these options in/etc/rc.conf. Contrast this with Debian where I have to edit/etc/init.d/sendmail. If a new version of sendmail comes out then I have to manually compare my version of/etc/init.d/sendmail with the package maintainers. I would rather not do that.
There are other options in/etc/rc.conf as well. Such as running xdm or not. Instead of one file Debian has a whole lot of other little config files. Such as the stuff in/etc/default.
While I am thinking about it another irritation I had with Debian was that if I installed a server Debian would insist on running the server. In particular Portmap and NFSD. I wanted these installed on my firewall as sometimes it was convinient to mount some nfs shares. But in general I did not want to run these services. Yet every time they were upgraded dpkg would try to restart them. Very annoying.
As I stated above I do use Debian on the desktop (as in right now). I don't mind as much that there are some mystery processes running and that the config files are a little harder to manage. I have console access I can tweek things if needed. On a server I want a default install that is ready to go and just works. For those cases I am useing OpenBSD. Ofcourse if I had a dual proc machine I would run Debian/Linux.
Re:Linux for desktop, *BSD for servers?
on
FreeBSD 4.6
·
· Score: 2
A 12MB Debian system is hardly useable.
As for bloat: exim requires libldap. That is just one example. The Debian config has also become a little top heavy.
My main Debain complaint is the lack of a good useable default install. Instead Debian is the "linux for everyone." This is an admirable goal, but not very usefull for servers.
Use OpenBSD for a few months and you will understand
Re:Linux for desktop, *BSD for servers?
on
FreeBSD 4.6
·
· Score: 5, Informative
I have used FreeBSD, OpenBSD, and various Linux distros on i386 hardware. In my experience Linux and FreeBSD are excellent on the desktop, and FreeBSD and OpenBSD are excellent on servers.
I find that the default install (without X) of both FreeBSD and OpenBSD has "everything I want in a server and nothing more." The ports system is there for the few extras you may want (like bash). Basically the defaults for the ports system and the install are sane. When I want a server I install *BSD get it running and forget about it. Usually I install OpenBSD as the install is easier, and it is slightly smaller.
No linux distro gives you this. I love Debian but it is suffereing from bloat. That and the default Debian install isen't good enough, because there isen't a true default. Something about giving the user choice. I don't need choice on my servers. I want an install that has been tested and works. Slackware dosen't have a ports/package system like FreeBSD. Again I want packages that has been tested and work. Slackware also has a hideous config. Editing all those files in/etc/rc.d/ is not the way to go. Use OpenBSD and you will understand. Gentoo is interesting. I just started useing it. It has promise, but it needs an easier install. It also needs a better default install. I like it and will continue to use it, but not on my servers. You really have to sit down and use FreeBSD or OpenBSD for a while before you will understand how lacking Linux distros are when it comes to servers.
For the desktop I have been useing Debian. I don't care too much about the bloat on the desktop and 'apt-get install package-name' is great. When it came to a desktop shoot out between Debian and FreeBSD, Debian/Linux won becuase ALSA supports my Trident 4D-NX sound card better then FreeBSD. In my experience Linux often supports uncommon bits of hardware better then FreeBSD. There isen't a native Mozilla for OpenBSD, so I haven't really used OpenBSD on the desktop.
For firewalls I have not used FreeBSD, only OpenBSD. OpenBSD has one of the best packet filters out there. It is easy to configure, and works. FreeBSD has something very similar. Recently I have been useing Linux as a firewall due to some funky stuff you can do with equalcost routeing, QOS, and bandwidth shapeing. If you don't need these features then OpenBSD is best. Linux can do some packet bashing that rivals Cisco routers. Unfortunately these features are largely undocumented.
Lack of documentation is ofcourse the worst part of Linux. FreeBSD and OpenBSD have lots documentation that is kept up to date. Linux dosen't.
FreeBSD and OpenBSD are better then any Linux distrobution for servers. These *BSD systems are well thought out and mature products. OpenBSD has a slight edge due to its easier install. On the desktop I think it is a tie. FreeBSD is excellent, but lacks a few of the bells and wistles you will find on a Linux destop. In particular some hardware is better supported under Linux. On the other hand Linux distors suffer from bloat and are not as well thought out as FreeBSD. OpenBSD makes an excellent firewall. Linux makes a good hybrid firewall/router. If I had to choose just one I would install FreeBSD everywhere.
Switch to grub. It is an advanced boot loader. Grub understands the linux filesystems so it will find your kernel at boot time. Unlike lilo grub dosen't have to be run every time you install a new kernel. Grub has a command line interface so you can easily load any kernel on your hard drive.
I see no reason to continue to use lilo. Except if you are in a very tight disk space situation such as embeded linux. In these cases every byte countes, and the ~150kB of grub might be a problem.
I had chills as Joda drew his sabre. In that instant I was thinking, "Excellent. Now we will see *The* Master Jedi in action." Unfortunately my ephoria was quickly dashed aside.
I was hopeing to see a grand master swordman. Where every parry, and stroke were perfect. Instead we get to see a buzzing bee zip around the screen.
In North America, one of the reasons that the Psion 5 is not more popular is because it is damn hard to get. When ever I have seen one, it was always a back ordered or special ordered item. Contrast this with the palms which are spammed across most electronics stores.
Granted the Revo and Diamond branded PDA is easier to get then the 5mx. But Revo is lacking compared to its bigger brother.
on a related note: I worked in a computer manufactureing and assemply plant. Naturally the plant runs 24/7, and people share equipment and workspaces across shifts. There was rarely a problem with equipment cleanliness. Except with the microscopes used for precision soldering. In particular people would forget to sterilize the eye pieces before and after their shift. This resulted in the occational massive outbraek of "pink eye." An annoying, and very contagious eye infection. You only need to get pink eye once to remember to clean the eye pieces with plenty of alcohol everytime you sit back down at a scope.
Two many people know how to write code, but don't know how to develope code. A symptom of this problem is that you get code that was written on a Linux platform, but was not written or designed for a Unix platform. So yes there is lots of unportable code.
The need for a ports system would exist regardless of the "quality of free software." Perfectly portable code will always require some patching to configure the software for the target OS. Even if all the patch does is move the location of the documents from/usr/share/doc to/usr/doc. Until every distro, and every Unix-like OS is identical this will be the case.
Additionally make sure your solder is NO CLEAN. Mixing NO CLEAN and other types of flux create a sticky, gummy, impossible to clean mess (which will also destroy the board).
Slashdot Mirror
In is not just the new drivers (teens). A large number of drivers are really unsafe. In North America we treat driving as a right instead of the privilege it is.
There is a big difference in having a 650 MB CDR sitting in your desk and 100 MB of storage space sitting up on an Apple server.
You are correct. The CDR is larger and more reliable.
If you had a fire? Theft? Water damage? All your computers (and disks) go with it. If Apple is hosting your back up, you can get it back.
This is silly. For a couple bucks a year I can cram a lot of offsite CDR backups in my safety deposit box. Failing that I can swap offsite storage with my buddy.
Any software, Sun's or MS's, gets insanely expensive in a hurry. If you took a 1000 person company and compared the true Total Cost of Ownership (TCO) of a complete Solaris system vs. a complete Win system you would probably find that the cost is the same. A company of this size should have support contracts for the sofware and contracts for hardware, plus a small army of admins. After those yearly costs the initial hardware and licence fees are peanuts. Even if the initial licence fee is $0 for a free Unix. Both Sun and MS have tried to prove that the TCO is in their favour. Neither have been particularly convinceing. You can easily replace "Sun" with any Unix vendor.
All vendors use per seat licenseing to some degree or another. You might be paying for a per seat license directly ala MS, or you might pay in upfront costs ala Sun. The per seat licenseing really only affects the little guys. Most of the bigger sites will just buy a site wide license if everyone is going to use the software. It should be noted that most of the interesting software is licensed per seat. All Sun really ships that is not per seat licenced is a fileserver and print spooler. If you use a free Unix you get a "complete exterprise solution" for free provided you have a small army of admin staff. This is were TCO kicks in. Licenseing is only a small part of TCO. It is the support contracts and admin staff that will really determine the TCO.
For any system that combines a mix of platforms the TCO will usually be higher. This is all this MS page really shows. A windows desktop is assumed.
Today due in large part to StarOffice/OpenOffice, Mozilla, and Evolution any office of any size could drop MS from the desktop and be quite happy. However the cost of switching remains really really high.
The scary part is that if you start spamming your link everywhere you will get donations.
I agree. The whole intimidation thing is real kicker isen't it. The people I hate are the ones who want to be intimidated. They feel superior for being stupid. How does that work? Compare this to my grandfather who seems to do alright despite being half blind[1]. Due to his handicap I think he looks at anything new as a challenge, and plesantly surprised when it isen't. Other people are probably pissed that a computer is supposed to be easy to use, as per marketing, but isen't. It makes them feel stupid. Those of us with more experience aren't intimidated, and when a computer dosen't work we think the developers are stupid. Maybe this is what we should tell users?
Instead of trying to convince users that computers are easy to use do the opposite. Tell them that computers are a pain in the ass (the truth). Tell users that the user interface was "designed" by a bunch stupid punk assed condesending arrogant new hires whacked out on caffine (possibly true). Now when the user manages to remember the basics they will fell a sense of superiority and acomplishment. I am only half kidding.
Regardless of the true complexity I often start off a mini training session with "I don't know why they had to make this so complicated..." It works for some people. Others are really stupid.
[1] Setting up a computer for my grandfather is what started me thinking about how poor computer user interfaces really are. My grandfather has lost much of his vision, but can still read if the text is magnified to atleast 3cm/inch high letters. A typewrite has a small font. There is not much you can do about it. In theory a computer should be able to display any sized text even if there is only one letter per screen. Too bad theory dosen't work out. For many reasons it is much easier for him to use a simple magnifying lens then to try to fight with the OS and display big fonts. How peverted is that?
Part of the problem there is that the users have probably had bad experiences with manuals in the past. Most computer manuals I have tried to read are very poorly written. I can't blame a user for looking at a computer manual in disgust, as I do the same. In general, documentation abuses the language, is poorly organized, and incomplete.
The computer industry does not give the users any credit and assumes all users are idiots. This mindset has UI designers convinved that they have to map computers onto something familar (desktops). This makes as much sense as putting a stick shift on a toaster to control the darkness. It will work, but is it really the best interface? By attempting to map "familiarity" the designers only frustrate the users when the familiarity inevetibly breaks due to a bug or a poor metaphor. Nothing frustrates a user more then when the actual outcome is different from the expected outcome. You can't map a car's interface onto a computer. Nor can you compared a car's "easy" interface to a computer's. A computer is a completely different device.
Here lies the problem: instead of treating computers as the powerfull devices they are, designers instead attempt to "perfect" the poor metaphors adopted years ago. Computer interfaces could be far more powerfull then they are. Instead a computers spends most of its power by drawing pretty pictures. Computer interfaces should expose the user to the power in the computer. This dosen't mean tab upon tab of check box options. Designers have to find new ways to present data. Not pretty ways, but usefull ways.
Not that this will happen. It is much easier to sell an old program with a new look then it is to discover better ways to manipulate data.
100 degrees in 2pi radians
400 actually, and the units are called gradians.
far to entrenched
The answer is to make both keys Control. Then maybe make caps lock alt-numlock or something equally weird.
Except in this case the vendors, such as Debian, didn't have any head start. As per the above post Theo didn't tell the vendors anything either.
There is a chance that your passphrase could end up in the swap file. Use an encrypted swap to be safe.
IIRC FreeBSD did this as well, during the install
/etc/rc.conf file. It is a file of sh parameters that are sourced by the rc scripts at boot time. In it you have config options for most of the major services that you would run. For example if I want sendmail to run with "-q30m -bd" (run queue every 30mins, listen on port 25) I set these options in /etc/rc.conf. Contrast this with Debian where I have to edit /etc/init.d/sendmail. If a new version of sendmail comes out then I have to manually compare my version of /etc/init.d/sendmail with the package maintainers. I would rather not do that.
/etc/rc.conf as well. Such as running xdm or not. Instead of one file Debian has a whole lot of other little config files. Such as the stuff in /etc/default.
True enough. I just did a FreeBSD install and it was a little different then I remember. I guess most of what I have written only applies to OpenBSD now.
Also what do you mean by top heavy concerning the config files?
One of the beauties of OpenBSD is the
There are other options in
While I am thinking about it another irritation I had with Debian was that if I installed a server Debian would insist on running the server. In particular Portmap and NFSD. I wanted these installed on my firewall as sometimes it was convinient to mount some nfs shares. But in general I did not want to run these services. Yet every time they were upgraded dpkg would try to restart them. Very annoying.
As I stated above I do use Debian on the desktop (as in right now). I don't mind as much that there are some mystery processes running and that the config files are a little harder to manage. I have console access I can tweek things if needed. On a server I want a default install that is ready to go and just works. For those cases I am useing OpenBSD. Ofcourse if I had a dual proc machine I would run Debian/Linux.
A 12MB Debian system is hardly useable.
As for bloat: exim requires libldap. That is just one example. The Debian config has also become a little top heavy.
My main Debain complaint is the lack of a good useable default install. Instead Debian is the "linux for everyone." This is an admirable goal, but not very usefull for servers.
Use OpenBSD for a few months and you will understand
I have used FreeBSD, OpenBSD, and various Linux distros on i386 hardware. In my experience Linux and FreeBSD are excellent on the desktop, and FreeBSD and OpenBSD are excellent on servers.
/etc/rc.d/ is not the way to go. Use OpenBSD and you will understand. Gentoo is interesting. I just started useing it. It has promise, but it needs an easier install. It also needs a better default install. I like it and will continue to use it, but not on my servers. You really have to sit down and use FreeBSD or OpenBSD for a while before you will understand how lacking Linux distros are when it comes to servers.
I find that the default install (without X) of both FreeBSD and OpenBSD has "everything I want in a server and nothing more." The ports system is there for the few extras you may want (like bash). Basically the defaults for the ports system and the install are sane. When I want a server I install *BSD get it running and forget about it. Usually I install OpenBSD as the install is easier, and it is slightly smaller.
No linux distro gives you this. I love Debian but it is suffereing from bloat. That and the default Debian install isen't good enough, because there isen't a true default. Something about giving the user choice. I don't need choice on my servers. I want an install that has been tested and works. Slackware dosen't have a ports/package system like FreeBSD. Again I want packages that has been tested and work. Slackware also has a hideous config. Editing all those files in
For the desktop I have been useing Debian. I don't care too much about the bloat on the desktop and 'apt-get install package-name' is great. When it came to a desktop shoot out between Debian and FreeBSD, Debian/Linux won becuase ALSA supports my Trident 4D-NX sound card better then FreeBSD. In my experience Linux often supports uncommon bits of hardware better then FreeBSD. There isen't a native Mozilla for OpenBSD, so I haven't really used OpenBSD on the desktop.
For firewalls I have not used FreeBSD, only OpenBSD. OpenBSD has one of the best packet filters out there. It is easy to configure, and works. FreeBSD has something very similar. Recently I have been useing Linux as a firewall due to some funky stuff you can do with equalcost routeing, QOS, and bandwidth shapeing. If you don't need these features then OpenBSD is best. Linux can do some packet bashing that rivals Cisco routers. Unfortunately these features are largely undocumented.
Lack of documentation is ofcourse the worst part of Linux. FreeBSD and OpenBSD have lots documentation that is kept up to date. Linux dosen't.
FreeBSD and OpenBSD are better then any Linux distrobution for servers. These *BSD systems are well thought out and mature products. OpenBSD has a slight edge due to its easier install. On the desktop I think it is a tie. FreeBSD is excellent, but lacks a few of the bells and wistles you will find on a Linux destop. In particular some hardware is better supported under Linux. On the other hand Linux distors suffer from bloat and are not as well thought out as FreeBSD. OpenBSD makes an excellent firewall. Linux makes a good hybrid firewall/router. If I had to choose just one I would install FreeBSD everywhere.
Switch to grub. It is an advanced boot loader. Grub understands the linux filesystems so it will find your kernel at boot time. Unlike lilo grub dosen't have to be run every time you install a new kernel. Grub has a command line interface so you can easily load any kernel on your hard drive.
I see no reason to continue to use lilo. Except if you are in a very tight disk space situation such as embeded linux. In these cases every byte countes, and the ~150kB of grub might be a problem.
The "spare cycles" idea was probably scraped due to copyright infringement. See Dan Simmons Hyperion series. In particular the second book.
ok ok, "perfect" wasn't the best word to use. I toyed with the idea of useing "amazingly great" but it didn't seem to fit. :-)
Lucas has managed to avoid giving us any heroes we could enjoy watching.
Worst still, the only character worth watching is the evil protagonist, Papaltine.
Agreed.
I had chills as Joda drew his sabre. In that instant I was thinking, "Excellent. Now we will see *The* Master Jedi in action." Unfortunately my ephoria was quickly dashed aside.
I was hopeing to see a grand master swordman. Where every parry, and stroke were perfect. Instead we get to see a buzzing bee zip around the screen.
What happens when you press the F1 key?
When I press the F1 key this application pops up that almost, but never completely, answers my question.
In North America, one of the reasons that the Psion 5 is not more popular is because it is damn hard to get. When ever I have seen one, it was always a back ordered or special ordered item. Contrast this with the palms which are spammed across most electronics stores.
Granted the Revo and Diamond branded PDA is easier to get then the 5mx. But Revo is lacking compared to its bigger brother.
on a related note: I worked in a computer manufactureing and assemply plant. Naturally the plant runs 24/7, and people share equipment and workspaces across shifts. There was rarely a problem with equipment cleanliness. Except with the microscopes used for precision soldering. In particular people would forget to sterilize the eye pieces before and after their shift. This resulted in the occational massive outbraek of "pink eye." An annoying, and very contagious eye infection. You only need to get pink eye once to remember to clean the eye pieces with plenty of alcohol everytime you sit back down at a scope.
What does PBUH mean?
I am guessing "Praise Be Unto Him."
damn, even after a preview:
s/Two/Too/
I suspect you are trolling. but...
/usr/share/doc to /usr/doc. Until every distro, and every Unix-like OS is identical this will be the case.
Two many people know how to write code, but don't know how to develope code. A symptom of this problem is that you get code that was written on a Linux platform, but was not written or designed for a Unix platform. So yes there is lots of unportable code.
The need for a ports system would exist regardless of the "quality of free software." Perfectly portable code will always require some patching to configure the software for the target OS. Even if all the patch does is move the location of the documents from
Additionally make sure your solder is NO CLEAN. Mixing NO CLEAN and other types of flux create a sticky, gummy, impossible to clean mess (which will also destroy the board).