$0.99 is far above my threshhold for a lossy, DRM-laden song. I realize that as long as Apple has to pay the record companies on the order of $0.70/song, the price will never become reasonable. Considering the low distribution cost to the record companies, they could sell these at half the cost and make a LOT more money -- on volume.
I don't care that much about the liner notes or album covers. However, the though of not having an uncompressed version of all my albums make me cringe. I would never pay money for music that has had lossy compression applied to it, not even if it is Ogg at the highest quality. Not $.99, not $.25, not $.10. If it's not at least CD quality, it's not worth paying for. I will rip my own and listen to the compressed music, but I have to have the original and I will pay for that, if it's reasonably priced, which it is not, currently.
And as long as it's not DRMed. Then I'll do without.
Hope it got better than it was in 2001. We had an installation of Oracle Parallel Server (on HP N-Class servers), and with Oracle engineers, and HP engineers, and our own DBA, we could never keep the damned thing up for more than two days running. A couple of hundred thousand dollars worth of licenses and consulting fees, and the single "temporary" instance running on a little A-Class ended up being our defacto production database. We never did get OPS to work reliably, but the little one did the job with never a hitch.
Hmmm? You still use telnet to get to your servers?
Are you under the impression that it is necessary to get an SSL certificate from a CA in order to use an SSL connection? Maybe that would be true if I *were* a business. If I just want to be able to connect to my firewall without someone sniffing my password, then there's no need for anything more than the cert I generate myself.
I connect to my servers and other machines using ssh, and transfer files that way too. It's not that I'm particularly secretive, it's just that it's safer that way.
There are other reasons for encryption than secrecy.
It is quite easy for an audit to find you in violation, whether you have "pirated" software or not. If you don't have a detailed audit trail for every license you own, if you can't lay your hands on the paper copy of every license with *no lapses*, you'll be nailed, even if you did in fact legally purchase each copy of every software package you have.
Most tightly run corporations would be at some risk for this. Most schools would not fall under the category of "tightly run", even if they worked very hard at being completely legal.
Strictly speaking, it is not necessary that *any* changes ever get submitted *back*. What is necessary is that the downstream "customer" must have the source code available to them. So if you make changes for the rest of your developer team, then all those developers must have access to the source code. If a student hacks a kernel and gives copies to his friends, those friends must have access to the source code.
On the other hand, if your modifications made it into your hardware product and you sell this product, your *customers* must have access to your modifications. Either that, or you must build your modifications in such a way that they are distinct from the GPL software you are working with, in which case you're under no obligation at all to provide source.
It doesn't stop responding here, it just stops logging. There was a time when it would go forever -- just don't expect it to restart when you made a config change and tried to restart to make the change "take". You could spend a frantic few minutes trying to track down why it wouldn't start again.
Haven't seen that problem in some time though. Now it will restart fine but give you no indication that that virtual host wasn't logging any more.
Just so we are clear you also shouldnt need to put in any effort after the install to lock down a system, it should be that way by default.
I agree. However, other than Gentoo, I've never seen a Linux distribution that passes that test either. Gentoo is only clean at install because you have to specifically install nearly every package yourself, otherwise I'm sure it would have tons of stuff running that you don't need.
After I do a bare-bones Redhat or Fedora install, I *do* have to go back through and disable all kinds of stuff. Debian is a lot better, but there is still stuff running after a bare bones install that needs to be cleaned out. It's been a few years since I installed Suse, so I can't speak to its level of required post install hardening. Solaris is probably as bad as RedHat in cruft that needs to be cleaned.
As I see it, where Unix has a clear advantage over Windows is in how much *easier* it is to track down unneeded services and get rid of them without bringing the system crashing down. Unix systems are very rarely secure when first installed, it's just easier to harden them.
A big factor here is that at some sites, the only way you can get a good idea of the price (and shipping charges, etc.) is to commit it to the shopping cart. I've even seen some sites that explicitly acknowlege that -- "To see the price put it in your shopping cart. You can delete it if you don't want it."
Thompson & Ritchie were handing out copies of the source code like candy in the early days. This was while AT&T was still under consent decree, and they were not allowed to be in the computer biz. Bell Labs was required by the terms of the consent decree to license its non-telephone technology to anyone who asked.
Haven't tried RAC, but we used Oracle Parallel Server a few years back. Hideously expensive, huge resource requirements...
And was far worse than any open source database with regards to jeopardizing our data. No end of problems. While we worked on it we kept a small server running an unlicensed standard Oracle instance, and it soldiered on without a lick of trouble, but OPS was a nightmare. The mailing lists and forums showed that we weren't alone.
So we simultaneously experienced the best and the worst that Oracle could provide. The big bucks that we spent on OPS was no guarantee of quality.
*On the server* unzip the OO package and open a command line sesssion. CD to the package directory and type 'setup -net' instead of 'setup'.
Go through the regular installation procedure.
Then, when you want to install it for a user, *as that user on the client machine* navigate to the installed OO directory (in Program Files or wherever you installed it.) Click on the setup program in the network installed directory (not the one in the distribution package.) It will offer you the choice of a full install and a workstation install (as I recall) -- choose the workstation install, it's the smaller of the two.
It just installs some personalization files and you're good to go.
And before you say, "Yeah, but what will they need to use in college?" consider what you used in college. Was there anything that OO.org in its current imperfect state could not handle perfectly well? Typing essays and reports? Including a simple table or chart of your chem lab results?
I used a typewriter in college, you insensitive clod!
I've never seen a Thinkpad older than a year that could even get an hour. This T23 gets a good solid 19 minutes from bootup with full battery to "You have 4 minutes power remaining." On a good day. The replacement battery was worse. My old 390 won't even boot from battery. Back when I supported Thinkpads, none of the older ones would hold a charge for any length of time.
I've looked at the NCharge batteries, and I may consider one. I need something like that. But my preliminary impression was the same as the reviewer's -- that the form factor sucked. There are competitors that make a nicer version, but that are a LOT more expensive.
There's a big gap between expecting them to open source their software and asking them to refrain from suing anyone who copies their ideas to save someone's life.
There may be accountabililty issues with open sourcing life critical software. I'm skeptical, because I think the public review does more to ensure safe code than strict accountability does, but that's an arguable question.
But suing someone who writes software using a similar approach in order to save lives is just wrong.
Not the newest, but who cares? Security fixes are backported to those versions. If you're running a server, that's what you want.
And Woody is due to be end-of-lifed here pretty soon. This is where you will have to deal with that "occasional version bump" which can pretty much be done in-line without a ton of disruption. You'll then be using what will be the new stable version, Sarge, which currently is using apache 2.0.52 (or 1.3.33, your choice), and PHP 4.3.10.
The versions of the packages are irrelelvant. The only time you have to deal with version changes is when you change from one version of the distribution to another.
For a desktop I want newer applications. For a server, I want it to work and to continue to work.
Actually, yes. Applying patches is reasonably quick. Compiling an updated package is usually not terribly time consuming, and the old package is still operational up until the moment the new package is installed.
Unless your server has a *very* heavy cpu load all the time, compiling source versus loading binaries isn't much of an issue.
I updated my workstation this morning -- about half a week's worth of updates. It took 18 minutes, all in the background.
I like Gentoo. All my home machines and one of my work machines runs on Gentoo. Nor would I dismiss out of hand the idea of running Gentoo on a server.
However...
As an administrator, I'm not particularly intrested in a distribution that will "teach me the inner workings of Linux". Stability and predictability are lots more important for production machines.
The new servers I'm putting on line now are all running Debian, and I'll be switching some old RH9 servers to Debian as I get the time to do that.
Someone earlier emphasized package management as a prime requirement for easy administration. Debian does that very well. Gentoo is also pretty good, except when things break, which does happen. I see Debian as more stable, Gentoo as more configurable. For a desktop, I'd choose Gentoo, but so far I'm leaning to Debian in the server room.
My biggest objection to most of the commercial distributions is that they are far too "versioned". If old versions had security updates forever, that'd be fine, but having to do a disruptive upgrade every few years on running servers just because there are no more security updates on the running version is quite inconvenient. This is one place where Gentoo really shines, being essentially "versionless". Debian makes version shifting relatively simple, so I'm comfortable with the relatively infrequent version bumps I'm likely to see.
How many programs implement just one patent (expired or not)?
The nature of software is that any given program (even a simple one) likely implements dozens of patents without even realizing it. The patents are not necessarily the central basis for the application, but inherent in the nuts and bolts of putting it together.
Slashdot Mirror
$0.99 is far above my threshhold for a lossy, DRM-laden song. I realize that as long as Apple has to pay the record companies on the order of $0.70/song, the price will never become reasonable. Considering the low distribution cost to the record companies, they could sell these at half the cost and make a LOT more money -- on volume.
But that's not gonna happen.
Because Campbell's Soup was happy with the free advertising.
But that was then, this is now. I'll bet Campbell's would be very upset if he were to paint those today.
I don't care that much about the liner notes or album covers. However, the though of not having an uncompressed version of all my albums make me cringe. I would never pay money for music that has had lossy compression applied to it, not even if it is Ogg at the highest quality. Not $.99, not $.25, not $.10. If it's not at least CD quality, it's not worth paying for. I will rip my own and listen to the compressed music, but I have to have the original and I will pay for that, if it's reasonably priced, which it is not, currently.
And as long as it's not DRMed. Then I'll do without.
Hope it got better than it was in 2001. We had an installation of Oracle Parallel Server (on HP N-Class servers), and with Oracle engineers, and HP engineers, and our own DBA, we could never keep the damned thing up for more than two days running. A couple of hundred thousand dollars worth of licenses and consulting fees, and the single "temporary" instance running on a little A-Class ended up being our defacto production database. We never did get OPS to work reliably, but the little one did the job with never a hitch.
It's not. And the commerical skip feature is why Sonic Blue no longer owns ReplayTV. They got spanked.
Hmmm? You still use telnet to get to your servers?
Are you under the impression that it is necessary to get an SSL certificate from a CA in order to use an SSL connection? Maybe that would be true if I *were* a business. If I just want to be able to connect to my firewall without someone sniffing my password, then there's no need for anything more than the cert I generate myself.
I connect to my servers and other machines using ssh, and transfer files that way too. It's not that I'm particularly secretive, it's just that it's safer that way.
There are other reasons for encryption than secrecy.
It is quite easy for an audit to find you in violation, whether you have "pirated" software or not. If you don't have a detailed audit trail for every license you own, if you can't lay your hands on the paper copy of every license with *no lapses*, you'll be nailed, even if you did in fact legally purchase each copy of every software package you have.
Most tightly run corporations would be at some risk for this. Most schools would not fall under the category of "tightly run", even if they worked very hard at being completely legal.
Strictly speaking, it is not necessary that *any* changes ever get submitted *back*. What is necessary is that the downstream "customer" must have the source code available to them. So if you make changes for the rest of your developer team, then all those developers must have access to the source code. If a student hacks a kernel and gives copies to his friends, those friends must have access to the source code.
On the other hand, if your modifications made it into your hardware product and you sell this product, your *customers* must have access to your modifications. Either that, or you must build your modifications in such a way that they are distinct from the GPL software you are working with, in which case you're under no obligation at all to provide source.
That's right! Who the hell is this Bob Metcalf guy, and why does he think he knows anything about computers!?
Ohh. That Bob Metcalf...
It doesn't stop responding here, it just stops logging. There was a time when it would go forever -- just don't expect it to restart when you made a config change and tried to restart to make the change "take". You could spend a frantic few minutes trying to track down why it wouldn't start again.
Haven't seen that problem in some time though. Now it will restart fine but give you no indication that that virtual host wasn't logging any more.
Just so we are clear you also shouldnt need to put in any effort after the install to lock down a system, it should be that way by default.
I agree. However, other than Gentoo, I've never seen a Linux distribution that passes that test either. Gentoo is only clean at install because you have to specifically install nearly every package yourself, otherwise I'm sure it would have tons of stuff running that you don't need.
After I do a bare-bones Redhat or Fedora install, I *do* have to go back through and disable all kinds of stuff. Debian is a lot better, but there is still stuff running after a bare bones install that needs to be cleaned out. It's been a few years since I installed Suse, so I can't speak to its level of required post install hardening. Solaris is probably as bad as RedHat in cruft that needs to be cleaned.
As I see it, where Unix has a clear advantage over Windows is in how much *easier* it is to track down unneeded services and get rid of them without bringing the system crashing down. Unix systems are very rarely secure when first installed, it's just easier to harden them.
That was a cover. Originally an early 70s song (late 60s?). Steam, I think.
A big factor here is that at some sites, the only way you can get a good idea of the price (and shipping charges, etc.) is to commit it to the shopping cart. I've even seen some sites that explicitly acknowlege that -- "To see the price put it in your shopping cart. You can delete it if you don't want it."
Thompson & Ritchie were handing out copies of the source code like candy in the early days. This was while AT&T was still under consent decree, and they were not allowed to be in the computer biz. Bell Labs was required by the terms of the consent decree to license its non-telephone technology to anyone who asked.
Not public domain, but pretty danged open.
Haven't tried RAC, but we used Oracle Parallel Server a few years back. Hideously expensive, huge resource requirements...
And was far worse than any open source database with regards to jeopardizing our data. No end of problems. While we worked on it we kept a small server running an unlicensed standard Oracle instance, and it soldiered on without a lick of trouble, but OPS was a nightmare. The mailing lists and forums showed that we weren't alone.
So we simultaneously experienced the best and the worst that Oracle could provide. The big bucks that we spent on OPS was no guarantee of quality.
*On the server* unzip the OO package and open a command line sesssion. CD to the package directory and type 'setup -net' instead of 'setup'.
Go through the regular installation procedure.
Then, when you want to install it for a user, *as that user on the client machine* navigate to the installed OO directory (in Program Files or wherever you installed it.) Click on the setup program in the network installed directory (not the one in the distribution package.) It will offer you the choice of a full install and a workstation install (as I recall) -- choose the workstation install, it's the smaller of the two.
It just installs some personalization files and you're good to go.
And before you say, "Yeah, but what will they need to use in college?" consider what you used in college. Was there anything that OO.org in its current imperfect state could not handle perfectly well? Typing essays and reports? Including a simple table or chart of your chem lab results?
I used a typewriter in college, you insensitive clod!
I always wanted to say that...
Ummm. That isn't the interface they're talking about?
I've never used anything but the web interface. Didn't know there was something else.
Wow! Has IBM finally figured out batteries?
How old is it?
I've never seen a Thinkpad older than a year that could even get an hour. This T23 gets a good solid 19 minutes from bootup with full battery to "You have 4 minutes power remaining." On a good day. The replacement battery was worse. My old 390 won't even boot from battery. Back when I supported Thinkpads, none of the older ones would hold a charge for any length of time.
I've looked at the NCharge batteries, and I may consider one. I need something like that. But my preliminary impression was the same as the reviewer's -- that the form factor sucked. There are competitors that make a nicer version, but that are a LOT more expensive.
There's a big gap between expecting them to open source their software and asking them to refrain from suing anyone who copies their ideas to save someone's life.
There may be accountabililty issues with open sourcing life critical software. I'm skeptical, because I think the public review does more to ensure safe code than strict accountability does, but that's an arguable question.
But suing someone who writes software using a similar approach in order to save lives is just wrong.
Ummm... That's make it uncommon, wouldn't it?
??
What version of Debian are you looking at?
My old Woody box has Apache 1.3.26 and PHP 4.1.2.
Not the newest, but who cares? Security fixes are backported to those versions. If you're running a server, that's what you want.
And Woody is due to be end-of-lifed here pretty soon. This is where you will have to deal with that "occasional version bump" which can pretty much be done in-line without a ton of disruption. You'll then be using what will be the new stable version, Sarge, which currently is using apache 2.0.52 (or 1.3.33, your choice), and PHP 4.3.10.
The versions of the packages are irrelelvant. The only time you have to deal with version changes is when you change from one version of the distribution to another.
For a desktop I want newer applications. For a server, I want it to work and to continue to work.
Actually, yes. Applying patches is reasonably quick. Compiling an updated package is usually not terribly time consuming, and the old package is still operational up until the moment the new package is installed.
Unless your server has a *very* heavy cpu load all the time, compiling source versus loading binaries isn't much of an issue.
I updated my workstation this morning -- about half a week's worth of updates. It took 18 minutes, all in the background.
I like Gentoo. All my home machines and one of my work machines runs on Gentoo. Nor would I dismiss out of hand the idea of running Gentoo on a server.
However...
As an administrator, I'm not particularly intrested in a distribution that will "teach me the inner workings of Linux". Stability and predictability are lots more important for production machines.
The new servers I'm putting on line now are all running Debian, and I'll be switching some old RH9 servers to Debian as I get the time to do that.
Someone earlier emphasized package management as a prime requirement for easy administration. Debian does that very well. Gentoo is also pretty good, except when things break, which does happen. I see Debian as more stable, Gentoo as more configurable. For a desktop, I'd choose Gentoo, but so far I'm leaning to Debian in the server room.
My biggest objection to most of the commercial distributions is that they are far too "versioned". If old versions had security updates forever, that'd be fine, but having to do a disruptive upgrade every few years on running servers just because there are no more security updates on the running version is quite inconvenient. This is one place where Gentoo really shines, being essentially "versionless". Debian makes version shifting relatively simple, so I'm comfortable with the relatively infrequent version bumps I'm likely to see.
How many programs implement just one patent (expired or not)?
The nature of software is that any given program (even a simple one) likely implements dozens of patents without even realizing it. The patents are not necessarily the central basis for the application, but inherent in the nuts and bolts of putting it together.