Mmm. Yes, I was thinking the same thing. The Mozilla guys decided to do a complete rewrite to nail IE into the ground...and look what happened to them (not talking about product quality, just market share).
Just throw books away when you're done with them! Keeps the economy moving!
If you really like a book, you'll get copies. I read Snow Crash in a library, I'm sure that any decent library would have it, but I still bought three copies of the book.
Heck... sash -ln -s/lib/i686/libc-2.2.4.so/lib/i686/libc.so.6 would have fixed this on my system.
If he'd already rebooted the system so that it couldn't boot, then he'd have a bit more of a problem. He'd have to make the drive and then type the following: lilo: linux init=/sbin/sash -ln -s/lib/i686/libc-2.2.4.so/lib/i686/libc.so.6
Voila. And I don't consider myself much of a UNIX guru.
I think it's less that Windows admins aren't any good (the impression I got from the original poster) and more than the really good sysadmins often migrate to UNIX. Obviously there are also good Windows sysadmins -- I've just found that the general level of competence among UNIX sysadmins is higher.
And frankly (and this could be because I don't know Windows as well), I'm not sure what I'd do to fix a situation in which something in the registry on a Windows box had gotten screwed up, preventing booting. I *could* handle the problem if it were initscripts on a UNIX box, however. In Windows, I'd probably be forced to restore from a backed up registry or reinstall everything after grabbing what I could from the hard drive by tossing it into another machine (I couldn't even boot into Windows from a CD because Windows is pretty much incapable of booting from a CD).
Almost anything that breaks on a UNIX box I can fix or can figure out how to track down and fix. Why? I have a really top-notch set of tools that make it easy: strace, ldd, ltrace, gdb with core files, rpm -V, lsof, netstat -nta,/var/log/messages...an average Linux system has an incredibly powerful and good set of troubleshooting tools that'll let almost anything be patched up. On Windows, almost nothing (well, checkdisk and regedit) is bundled that lets you analyze and fix problems. There are a very few good free utilities out there -- filemon and regmon, but on the whole, fixing Windows problems is as much fun as banging your head into a wall. Hell, Microsoft doesn't even put out a free debugger -- even Apple (which seems to be much derided by tech enthusiasts) gives away the solid Macsbug and ResEdit...I don't see MS handing out SoftICE and Borland Resource Workshop any time soon.
Also, I've seen more availability issues with Windows than UNIX. At CMU, we have a nice collection of big UNIX machines we can use -- they're simply up. You can log into them, and they always work.
The Windows machines in clusters around campus are another story. One day your AFS directory doesn't mount, another the central auth server isn't handing back anything to the Windows clients.
Finally, I think you're somewhat guilty of the "reboot" mentality as well. Okay, maybe the reason the guy didn't want to reboot was bogus, but there's no reason to reboot. I don't even know why he wanted to kill the servers --/etc/rc.d/init.d/servicename reload or kill -HUP should, for any demon, produce a reload of the config files without restarting. The only time you'd want to actually *kill* any daemon is when you've actually replaced the binary with a new version, and you want to load that into memory so that it's being used.
I also don't see why, if you were working on half of a load balanced-site, you couldn't redirect the work to the other system for a while and reboot when your server was unused, if you were bound and determined to reboot.
Um...at least zsh does the cycling that you want. That kind of cycling drives me mad, though my roommate loves it, so YMMV. Frankly, cmd.exe is awful -- if you know what you're doing, it's just a waste of time. With a huge amount of work (add cygwin, flip on tab completion, add a few native ports of gnu programs), you can actually get the thing approaching the usability of a very slow UNIX box. If you're going to that much effort to try to get a usable command line environment, you should probably be using Linux anyway.
Besides, bash is available for Windows.
Not having emacs-style readline editing would tick me off.
IIRC, cmd.exe doesn't do all the neat history operations:
./configure && nice -20 make
!./co and you've just rerun the command
how about ls car*.jpg -lhS|grep ferrari|tr -s " "|cut "-d " -f5,9|head -n10 that'll list the ten largest images of Ferrari cars in your current directory along with their sizes. Now suppose you want to see your Porche images You *could* tap up, ctrl-a, hold the right arrow a while, delete the ferrari text and enter "porche". *or* you could just type ^ferrari^porche voila. cmd.exe lacks intelligent completition based on the command you're completing -- for example, completing a path on a cd command will only complete directories in zsh and bash (with bash-completition installed). On cmd.exe, no such luck.
I don't understand why people *care* unless they're writing something that's absolutely performance-critical (and then I suspect it'd be in perl, not in shell).
Yes, cat causes a bit of overhead. However (and I have a PII/266), it serves one excellent purpose -- if you habitually use cat, you don't have to worry about the syntax of the next command in the pipeline. Maybe it accepts a file, maybe it doesn't, maybe it's some wrapper script that doesn't accept a file...for one liners, using cat is a *good* habit to be in. You don't have to constantly check man pages, documentation, or "try it and see if it works".
I think most of the "useless use of" complaints in UNIX that USENET people like to mention to show off their UNIX leetness are just stupid. Skimming down this page, I see:
Useless Use of kill -9: Okay, you should always use TERM first. However, spending more of your valuable time trying three or four other signals before kill -9 is just stupid, however (unless it's netscape, which oddly enough can sometimes be axed with QUIT).
Useless use of echo may be more legitimate *but* you may be planning to do something more extensive: command -options `echo $variable|seds/foo/bar/`, for example. You can't do that with just command -options $variable
The useless use of ls * really is useless, IMHO, because it frequently has unwanted side effects...you'd need ls -d * to get equivalent behavior to ls.
They get cranky about using grep foo |wc -l instead of grep -c foo. Who *cares*? Frequently grep is not the last element in the pipeline. If you're in the habit of using wc -l, you don't have to worry about the preceding item, even if you decide to insert a new item into the pipeline before grep and wc.
The complaint about using grep + awk is just stupid. grep is significantly faster than awk, anyway.
Well, probably because id makes awful games. Oh, don't get me wrong -- Carmack makes incredible *engines*. But as for actual gameplay...Nobody buys Quake for the game. You buy it so you can play mods. id makes a new engine and then licenses it to all the people making Quake-engine games, and while they're doing that, they'd be silly to hand out the code for free.
I understand why she doesn't like the amount of time he spends on it, but I wanna know why she's so hung up about her boyfriend playing a female elf.
There shouldn't be a social stigma for every guy to play a male human and every female to play a female human. The entire point of these games is that you're role-playing something exotic, something not mundane. If you're just role-playing yourself, save the money you're sending to the EQ people and just go out and live life.
I've been really irritated by this for some time. Okay, I can vaguely understand blocking port 22 inbound for security reasons. You don't want any chance of someone getting into a poorly set up system that's also running ssh.
But why do employers block 22 outbound? It's really annoying. You're more likely to use ssh for work than any other protocol I can think of.
A guy I know worked at a site where ssh outbound was blocked, but telnet outbound was allowed (he ended up running ssh on the telnet port on his box to avoid this). I just don't see the logic behind these decisions -- how is allowing telnet but disallowing ssh security-tightening?
I see the practical value of firewalls -- most systems (esp. Windows boxes) are a PITA to administer remotely and en mass. I still don't like them.
Oh, christ. No one at work is going to be setting up elaborate forwarding systems for man-in-the-middle attacks. IT people do not have time for this kind of crap -- they have work to do. Running a packet logger is one thing, this entirely different
Now, perhaps it means that Jabber isn't the best thing to, say, use as a defense communication network, but it's a lot better than anything else out there. I'd feel comfortable using it to send critical business messages.
SSL is kind of overrated (good, and I happily SSL tunnel what I can, just the need for it is overrated). People agonize about submitting semi-sensitive forms in a web browser over a non-SSL connection. No one steals credit cards by packet sniffing. Much easier and more profitable in almost more cases to find someone that hasn't secured their CC/customer database. Packet sniffing is, in almost all real-world issues, limited to plain text attacks. It's really entertaining to see Bob down the hall in a cybersex MUSH or Mike sending long personal unencrypted emails, maybe, but unless you're a hardened hacker that's absolutely determined to get some information or into some system and there's *no* alternative, you're not going to bother with anything but plaintext.
Besides, Jabber can PGP-sign and encrypt data. That's more than enough for me, since I trust PGP more than I do SSL.
WV sucks -- the mandantory facial recognition. The Microsoft grant to the state board of education that's made IE mandantory, a Microsoft-chosen person responsible for determining tech education in schools, and student surveys on owned products in the home (which go to MS marketing research) mandantory in public schools. WV has the lowest literacy levels, the highest cancer rate, the lowest average income, and almost the highest age of any state in the United States.
The same logic *does* apply to a 1984-style scenerio, you realize. "Why don't you want the government to see everything in your life if you aren't doing anything wrong?" Yeah.
The reason Castro took over so easily in Cuba and squashed resistance is because all guns were registered. When he wanted to ensure that only the military had guns after the revolution, he just had soldiers go door to door. You'd better hope that you could cough up all the guns that you had listed as registered to you.
Agreed. Assuming the guy even *has* the ability to enter the data without a valid thumb scan, he isn't going to risk getting fired over you not wanting your thumb scanned. You'd have to go to a higher-up policymaker and make waves.
No need for new projects -- already good distributed filesystems that you can set up big servers with afs? (or here) coda? intermezzo?
CMU, for example, uses AFS campus-wide. Your login scripts and dotfiles and whatnot all reside in your home directory (on AFS) so preferences migrate with you.
You can make things world-readable, and because AFS has a global namespace, anyone can see them. If I do research at MIT as well, I just need to grab a Kerberos ticket from their KDC and start using my files over there.
Just plonk a server in place, put an array of 100GB drives in place, make things readable by whomever you want, and you're good to go.
If you want a system designed with fancy automated caching that people can use without dicking around with Kerberos, freenet's a good choice. Of course, there's no guarantee that the data will stay around, but cest la vie.
The x86 is a fairly ugly and lame architecture for just about anything these days.
It's a testament to the skill of the engineers at Intel and AMD that we actually have well-performing x86 chips -- of course, they use more power than the lamp on my desk...
I wish something like the Alpha (RIP) or the PowerPC or some new clean architecture would come around.
Be cool if IA-64 had thrown out IA-32 compatibility.
I hate to say it, but it would probably be a good move in terms of establishing prescedent for the EFF/FSF to sue the Lindows people.
It sucks from a real world perspective, given what the Lindows people are trying to do, but frankly, distributing a beta is still distributing binaries, and must be accompanied by source. To do otherwise and have it slide would open the GPL to abuse down the line.
I don't see why the Lindows people don't want to release source. Hand it out to their beta testers. No biggie. Snapshot their CVS.
Slashdot Mirror
Mmm. Yes, I was thinking the same thing. The Mozilla guys decided to do a complete rewrite to nail IE into the ground...and look what happened to them (not talking about product quality, just market share).
It's going to be interesting, this IIS rewrite.
Just throw books away when you're done with them! Keeps the economy moving!
If you really like a book, you'll get copies. I read Snow Crash in a library, I'm sure that any decent library would have it, but I still bought three copies of the book.
He was forced to drive 12 hours round trip?
/lib/i686/libc-2.2.4.so /lib/i686/libc.so.6 would have fixed this on my system.
/lib/i686/libc-2.2.4.so /lib/i686/libc.so.6
/var/log/messages...an average Linux system has an incredibly powerful and good set of troubleshooting tools that'll let almost anything be patched up. On Windows, almost nothing (well, checkdisk and regedit) is bundled that lets you analyze and fix problems. There are a very few good free utilities out there -- filemon and regmon, but on the whole, fixing Windows problems is as much fun as banging your head into a wall. Hell, Microsoft doesn't even put out a free debugger -- even Apple (which seems to be much derided by tech enthusiasts) gives away the solid Macsbug and ResEdit...I don't see MS handing out SoftICE and Borland Resource Workshop any time soon.
/etc/rc.d/init.d/servicename reload or kill -HUP should, for any demon, produce a reload of the config files without restarting. The only time you'd want to actually *kill* any daemon is when you've actually replaced the binary with a new version, and you want to load that into memory so that it's being used.
Heck...
sash
-ln -s
If he'd already rebooted the system so that it couldn't boot, then he'd have a bit more of a problem. He'd have to make the drive and then type the following:
lilo: linux init=/sbin/sash
-ln -s
Voila. And I don't consider myself much of a UNIX guru.
I think it's less that Windows admins aren't any good (the impression I got from the original poster) and more than the really good sysadmins often migrate to UNIX. Obviously there are also good Windows sysadmins -- I've just found that the general level of competence among UNIX sysadmins is higher.
And frankly (and this could be because I don't know Windows as well), I'm not sure what I'd do to fix a situation in which something in the registry on a Windows box had gotten screwed up, preventing booting. I *could* handle the problem if it were initscripts on a UNIX box, however. In Windows, I'd probably be forced to restore from a backed up registry or reinstall everything after grabbing what I could from the hard drive by tossing it into another machine (I couldn't even boot into Windows from a CD because Windows is pretty much incapable of booting from a CD).
Almost anything that breaks on a UNIX box I can fix or can figure out how to track down and fix. Why? I have a really top-notch set of tools that make it easy: strace, ldd, ltrace, gdb with core files, rpm -V, lsof, netstat -nta,
Also, I've seen more availability issues with Windows than UNIX. At CMU, we have a nice collection of big UNIX machines we can use -- they're simply up. You can log into them, and they always work.
The Windows machines in clusters around campus are another story. One day your AFS directory doesn't mount, another the central auth server isn't handing back anything to the Windows clients.
Finally, I think you're somewhat guilty of the "reboot" mentality as well. Okay, maybe the reason the guy didn't want to reboot was bogus, but there's no reason to reboot. I don't even know why he wanted to kill the servers --
I also don't see why, if you were working on half of a load balanced-site, you couldn't redirect the work to the other system for a while and reboot when your server was unused, if you were bound and determined to reboot.
Um...at least zsh does the cycling that you want. That kind of cycling drives me mad, though my roommate loves it, so YMMV. Frankly, cmd.exe is awful -- if you know what you're doing, it's just a waste of time. With a huge amount of work (add cygwin, flip on tab completion, add a few native ports of gnu programs), you can actually get the thing approaching the usability of a very slow UNIX box. If you're going to that much effort to try to get a usable command line environment, you should probably be using Linux anyway.
Besides, bash is available for Windows.
Not having emacs-style readline editing would tick me off.
IIRC, cmd.exe doesn't do all the neat history operations:
./configure && nice -20 make
!./co
and you've just rerun the command
how about
ls car*.jpg -lhS|grep ferrari|tr -s " "|cut "-d " -f5,9|head -n10
that'll list the ten largest images of Ferrari cars in your current directory along with their sizes. Now suppose you want to see your Porche images You *could* tap up, ctrl-a, hold the right arrow a while, delete the ferrari text and enter "porche". *or* you could just type
^ferrari^porche
voila.
cmd.exe lacks intelligent completition based on the command you're completing -- for example, completing a path on a cd command will only complete directories in zsh and bash (with bash-completition installed). On cmd.exe, no such luck.
I don't understand why people *care* unless they're writing something that's absolutely performance-critical (and then I suspect it'd be in perl, not in shell).
Yes, cat causes a bit of overhead. However (and I have a PII/266), it serves one excellent purpose -- if you habitually use cat, you don't have to worry about the syntax of the next command in the pipeline. Maybe it accepts a file, maybe it doesn't, maybe it's some wrapper script that doesn't accept a file...for one liners, using cat is a *good* habit to be in. You don't have to constantly check man pages, documentation, or "try it and see if it works".
I think most of the "useless use of" complaints in UNIX that USENET people like to mention to show off their UNIX leetness are just stupid. Skimming down this page, I see:
Useless Use of kill -9: Okay, you should always use TERM first. However, spending more of your valuable time trying three or four other signals before kill -9 is just stupid, however (unless it's netscape, which oddly enough can sometimes be axed with QUIT).
Useless use of echo may be more legitimate *but* you may be planning to do something more extensive:
command -options `echo $variable|seds/foo/bar/`, for example. You can't do that with just
command -options $variable
The useless use of ls * really is useless, IMHO, because it frequently has unwanted side effects...you'd need ls -d * to get equivalent behavior to ls.
They get cranky about using grep foo |wc -l instead of grep -c foo. Who *cares*? Frequently grep is not the last element in the pipeline. If you're in the habit of using wc -l, you don't have to worry about the preceding item, even if you decide to insert a new item into the pipeline before grep and wc.
The complaint about using grep + awk is just stupid. grep is significantly faster than awk, anyway.
Well, probably because id makes awful games. Oh, don't get me wrong -- Carmack makes incredible *engines*. But as for actual gameplay...Nobody buys Quake for the game. You buy it so you can play mods. id makes a new engine and then licenses it to all the people making Quake-engine games, and while they're doing that, they'd be silly to hand out the code for free.
Mmmm....
:-)
And I was upset about PGP being discontinued. Start selling "PGP Gaming Edition". It'd revitalize PGP.
I understand why she doesn't like the amount of time he spends on it, but I wanna know why she's so hung up about her boyfriend playing a female elf.
There shouldn't be a social stigma for every guy to play a male human and every female to play a female human. The entire point of these games is that you're role-playing something exotic, something not mundane. If you're just role-playing yourself, save the money you're sending to the EQ people and just go out and live life.
Uh, huh. Can *you* do it?
I've been really irritated by this for some time. Okay, I can vaguely understand blocking port 22 inbound for security reasons. You don't want any chance of someone getting into a poorly set up system that's also running ssh.
But why do employers block 22 outbound? It's really annoying. You're more likely to use ssh for work than any other protocol I can think of.
A guy I know worked at a site where ssh outbound was blocked, but telnet outbound was allowed (he ended up running ssh on the telnet port on his box to avoid this). I just don't see the logic behind these decisions -- how is allowing telnet but disallowing ssh security-tightening?
I see the practical value of firewalls -- most systems (esp. Windows boxes) are a PITA to administer remotely and en mass. I still don't like them.
Oh, christ. No one at work is going to be setting up elaborate forwarding systems for man-in-the-middle attacks. IT people do not have time for this kind of crap -- they have work to do. Running a packet logger is one thing, this entirely different
Now, perhaps it means that Jabber isn't the best thing to, say, use as a defense communication network, but it's a lot better than anything else out there. I'd feel comfortable using it to send critical business messages.
SSL is kind of overrated (good, and I happily SSL tunnel what I can, just the need for it is overrated). People agonize about submitting semi-sensitive forms in a web browser over a non-SSL connection. No one steals credit cards by packet sniffing. Much easier and more profitable in almost more cases to find someone that hasn't secured their CC/customer database. Packet sniffing is, in almost all real-world issues, limited to plain text attacks. It's really entertaining to see Bob down the hall in a cybersex MUSH or Mike sending long personal unencrypted emails, maybe, but unless you're a hardened hacker that's absolutely determined to get some information or into some system and there's *no* alternative, you're not going to bother with anything but plaintext.
Besides, Jabber can PGP-sign and encrypt data. That's more than enough for me, since I trust PGP more than I do SSL.
Wow, I just commented on that.
WV sucks -- the mandantory facial recognition. The Microsoft grant to the state board of education that's made IE mandantory, a Microsoft-chosen person responsible for determining tech education in schools, and student surveys on owned products in the home (which go to MS marketing research) mandantory in public schools. WV has the lowest literacy levels, the highest cancer rate, the lowest average income, and almost the highest age of any state in the United States.
That is downright disturbing. I know that revision of history happens even in the US, but seeing it in action is really unpleasant.
The same logic *does* apply to a 1984-style scenerio, you realize. "Why don't you want the government to see everything in your life if you aren't doing anything wrong?" Yeah.
"memes are what scare me"
Suppress ideas!
The reason Castro took over so easily in Cuba and squashed resistance is because all guns were registered. When he wanted to ensure that only the military had guns after the revolution, he just had soldiers go door to door. You'd better hope that you could cough up all the guns that you had listed as registered to you.
Agreed. Assuming the guy even *has* the ability to enter the data without a valid thumb scan, he isn't going to risk getting fired over you not wanting your thumb scanned. You'd have to go to a higher-up policymaker and make waves.
You're saying that *England* is less intrusive than the US?
Well, most users seem to disagree ( :-( ) given what's happened to PGP.
Depressing.
No need for new projects -- already good distributed filesystems that you can set up big servers with
afs? (or here)
coda?
intermezzo?
CMU, for example, uses AFS campus-wide. Your login scripts and dotfiles and whatnot all reside in your home directory (on AFS) so preferences migrate with you.
You can make things world-readable, and because AFS has a global namespace, anyone can see them. If I do research at MIT as well, I just need to grab a Kerberos ticket from their KDC and start using my files over there.
Just plonk a server in place, put an array of 100GB drives in place, make things readable by whomever you want, and you're good to go.
If you want a system designed with fancy automated caching that people can use without dicking around with Kerberos, freenet's a good choice. Of course, there's no guarantee that the data will stay around, but cest la vie.
Which system is based on a storage mechanism that *will* fail, screwing over the system owner?
Xbox
The x86 is a fairly ugly and lame architecture for just about anything these days.
It's a testament to the skill of the engineers at Intel and AMD that we actually have well-performing x86 chips -- of course, they use more power than the lamp on my desk...
I wish something like the Alpha (RIP) or the PowerPC or some new clean architecture would come around.
Be cool if IA-64 had thrown out IA-32 compatibility.
Wanna bet whether your 20 gig internal hard drive or the other guy's 8 meg memory card will fail first?
And it's pretty easy to pick up a new memcard.
And you are a corporate buyer for a chain selling X-Boxes?
Frankly, I'd say that whatever experience he has has at least some value. What experience do you have to make your own claims?
I hate to say it, but it would probably be a good move in terms of establishing prescedent for the EFF/FSF to sue the Lindows people.
It sucks from a real world perspective, given what the Lindows people are trying to do, but frankly, distributing a beta is still distributing binaries, and must be accompanied by source. To do otherwise and have it slide would open the GPL to abuse down the line.
I don't see why the Lindows people don't want to release source. Hand it out to their beta testers. No biggie. Snapshot their CVS.