he goal was to compare the processes required to set up and administer various services in Windows 2003 Enterprise vs. Red Hat's and SuSE's boxed enterprise server NOSes. [...] Long story short, Windows came out on top by a huge margin in every field - ease, usability, intuitiveness, support, everything.
Actually, I believe you. Having tried pretty much every major Linux distribution, I have to say that the commercial distributions really suck shit in those areas.
Take installing software, for example. RedHat and SuSE are only now finally getting a clue and supporting something like APT. SLES 9 was so crap I ended up writing my own tools to index the RPMs available on the repository and let me grep for the ones I needed, and feed the URLs to ncftpget.
Or installing the OS. I'm running OpenSuSE on a server because SLES wouldn't install properly, the X drivers for S3 were broken I think, X just went into an infinite loop when the machine tried to boot.
Or documentation. Type man yast or man yast2 on SLES, and you used to get "No manual entry for yast2", now you get a useless 1-page summary of how to change the widget style or use it as a replacement for rpm -i. RedHat's just as bad, last time I had the misfortune of running RHEL none of their add-on utilities had man pages.
As for intuitiveness, no computer is intuitive, so obviously Windows will win there because it's what most people are used to.
I find it really interesting that the commercial "enterprise" distributions are so much worse than the free ones. I could understand if they were no better, but my experience is that the commercial distributions provide a massive value subtract. You'd think they could at least be as up-to-date, easy to admin and well documented as the infamously slow-to-release Debian.
I'm not sure if you're going to believe me, but I have a Windows 2000 Server running for about 2.5 years now, serving as a small ASP-driven web server.
So you're saying you haven't installed a service patch to your Windows 2003 box that required a reboot in 2.5 years? Care to post the web server address? I'm betting you won't dare.
And the irony is, by forcing the open source community to develop GCJ-based Java solutions, they're destroying the supposed raison d'etre of Java, to be able to take the same binary and run it anywhere.
You probably bought your reader after the new Memory Stick Pro was introduced.
Yes, the new post-Memory-Stick-Pro reader I have can read Memory Stick Pro. The point is I needed to buy new hardware, whereas I didn't have that problem with CompactFlash.
Yeah. First there was Memory Stick. Then came Memory Stick Pro, and my Memory Stick reader couldn't read it, which was a pain as my Dad's camera took Memory Stick Pro.
Then I got a phone with Memory Stick Duo, and guess what? Phone couldn't read Memory Stick or Memory Stick Pro, reader couldn't read Duo.
Then came the Memory Stick Pro Duo.
Meanwhile, my new Canon SLR can use the same CompactFlash cards I was using 4 years ago, and my new 2GB CF card works in my existing card reader.
Nothing is perfect, but having reliable source authentication (so that everyone can easily tell which emails are really from PayPal and which are from criminals pretending to be PayPal) would go a long way towards minimizing the problems caused by phishing.
The thing is, we already have S/MIME. PayPal and the banks could start using it tomorrow if they wanted to. Every Mac user, every Mozilla Thunderbird user, and so on, already has client support.
They don't bother to sign their messages because it doesn't cost them a significant amount if you get scammed.
If you're worried about mounting deficits, why not go after a few of the corporations that make billions of dollars in profits, but mysteriously pay no taxes?
For example, News Corporation (Rupert Murdoch, FOX) who pay 0-7% tax. Or Pepsico, who paid 0% tax in 2000, and still pay very little.
Yeah, I've found that when I listen to Sparks, Pandora can't seem to suggest anything else that's remotely appealing. Either Sparks can't be explained by Pandora's categories, or there's nothing else like 'em.
It's better than SLES 9 on the server I've been setting up, in that curses programs like YaST actually work properly without locking up, and the video drivers for Unichrome work.
I always hated YaST for package management, so the option of using APT is a big win. I haven't tried the ZenWorks or 'smart' package managers everyone else is talking about.
Basically, for a server it's OK. I'd rather be running Debian, but the software I need to run is built for SLES.
If the owner wants to write in VB6, or you're not prepared to drink the Microsoft Kool-Aid and require.NET and a 15MB download, you might want to consider RealBASIC.
It's almost totally compatible with VB6, and compiles to native code on Windows, Macintosh and Linux, so you'd actually have the option of supporting multiple platforms.
Well, I've had two answers to what dnl means, and they're both totally different. So much for self-explanatory config...
Shell scripting isn't hard when it's done appropriately. It becomes a pain when you want to do anything moderately complex, but at that point you should be using a more appropriate tool anyway.
The fact that there are plenty of suckers willing to pay some parasite thousands of dollars for a domain name doesn't change the essentially worthless nature of a domain name, and doesn't change the fact that the people who squat on domains they aren't using in order to try and extort money are lowlifes.
My rule is: Don't feed the parasites. Think of a new name. Remember that none of the biggest most successful Internet companies paid off cybersquatters for "prime virtual real estate" like search.com or books.com; they all went with obscure names nobody had registered like "Google", "ebay" or "Amazon".
In fact, paying big bucks for a domain name is a pretty good predictor of business failure, as far as I can see.
GPLv3 aims to prevent the use of DRM to protect code, to ensure that it remains open to modification. Imagine a device that ships with embedded GPL'd code, but uses a digital signature to verify that only "authorized" versions of the GPL'd code can run.
We don't have to imagine such a device, there's one in my living room.
TiVo take Linux and use it to create their platform. They release the code as required, but it's utterly useless, because I can't modify the boot sequence of my TiVo to (say) make it start sshd--it's all locked down with DRM built into the hardware.
I have a perfectly legitimate, legal reason to want to be able to modify the code: I'd like to be able to SSH in to my TiVo from a remote location and tell it to record something.
Thus TiVo have taken GPL code, and prohibited me from modifying it and using it myself on hardware I own. That's exactly what the GPL was supposed to prevent from happening.
[And as for "the market will sort it out" claims--show me the satellite or cable company that provides a DVR that's open and I'll switch. And no, MythTV isn't a sensible option, because that involves decoding the MPEG stream to analog video and then re-encoding it back to MPEG again.]
When you look at it this way, you can see why there are more than a few people around who think DRM is a good idea. More than that, it's a necessary idea. You basically can't do what they want to do -- manufacture content and sell it per-unit, as if they were Ford or GM -- without some control that keeps people from deflating the price back to its actual marginal cost of production and distribution (the "one more copy" cost).
Except you can, and people do it all the time. Just yesterday I bought a bunch of FLAC files from a music download service, and some CDs. I'm about to order a book. I subscribe to a newspaper online, and a magazine on paper.
DRM isn't about being able to sell intellectual property for a profit, it's about being able to jack the prices up to previously unsustainable levels.
Sendmail is pre-Internet. It was built to route mail between BITNET, UUCP, ARPAnet, JAnet, and so on, all of which had different e-mail syntax. That's why it has a big slow crufty macro engine that every message goes through, and that's why it rewrites the headers of e-mail passing through it. None of that is necessary or desirable these days. Most of sendmail's other problems, from lack of speed to poor security, flow from that initial design decision, so you really need to start again from scratch with a simple e-mail parser and build up from there.
smtpd_helo_required = no
smtpd_helo_restrictions =
strict_rfc821_envelopes = no
smtpd_recipient_restrictions = permit_mynetworks,reject_unauth_destination
smtp_sasl_auth_enable = no
smtpd_sasl_auth_enable = no
smtpd_use_tls = no
smtp_use_tls = no
I know which I'd rather edit. I mean, without looking at the manual, I've no idea what that dnl crap is about.
I really like the current look of Slashdot. What was the point in changing it? Just to change it?
No, the point was to change it just enough to annoy people, but not enough to allow any changes that might fix the basic design issues. Because, you know, it has to look like Slashdot. So the contest was to come up with new lipstick for the pig. When I read the list of restrictions and found out the judges were looking for something pretty much like what they already had, I lost all interest.
No offense to the design winner, but too often CSS styles websites just end up a bunch of gradient filled rounded corner boxes. Its like the CSS community thinks with one brain cell.
That's nothing to do with CSS. Gradients and rounded corners are a bit of a bitch to do in CSS, in fact. They just happen to be the trendy thing in web design at the moment.
Rounded corners are easier with tables than with CSS. (There, I admitted it.)
Kutaragi was demoted after being passed over for the role of CEO and, when former Sony Pictures head Howard Stringer assumed the position, the relationship between the content and technology divisions of Sony became even more intimate.
I think they could have phrased that more tastefully.
And what about the highest performance file systems (XFS, ReiserFS) which don't?
Getting good performance out of clustering and padding is hard. Implementing high performance filesystem-style clustering for the file writing code of an office suite is ridiculously overoptimizing the wrong piece of code.
Slashdot Mirror
Actually, I believe you. Having tried pretty much every major Linux distribution, I have to say that the commercial distributions really suck shit in those areas.
Take installing software, for example. RedHat and SuSE are only now finally getting a clue and supporting something like APT. SLES 9 was so crap I ended up writing my own tools to index the RPMs available on the repository and let me grep for the ones I needed, and feed the URLs to ncftpget.
Or installing the OS. I'm running OpenSuSE on a server because SLES wouldn't install properly, the X drivers for S3 were broken I think, X just went into an infinite loop when the machine tried to boot.
Or documentation. Type man yast or man yast2 on SLES, and you used to get "No manual entry for yast2", now you get a useless 1-page summary of how to change the widget style or use it as a replacement for rpm -i. RedHat's just as bad, last time I had the misfortune of running RHEL none of their add-on utilities had man pages.
As for intuitiveness, no computer is intuitive, so obviously Windows will win there because it's what most people are used to.
I find it really interesting that the commercial "enterprise" distributions are so much worse than the free ones. I could understand if they were no better, but my experience is that the commercial distributions provide a massive value subtract. You'd think they could at least be as up-to-date, easy to admin and well documented as the infamously slow-to-release Debian.
[Opinions mine, very definitely not IBM's.]
So you're saying you haven't installed a service patch to your Windows 2003 box that required a reboot in 2.5 years? Care to post the web server address? I'm betting you won't dare.
And the irony is, by forcing the open source community to develop GCJ-based Java solutions, they're destroying the supposed raison d'etre of Java, to be able to take the same binary and run it anywhere.
One, my CF card reader was old enough that it was USB 1.1.
Two, do the Sony readers work on Linux and Mac? If not, what use is that?
You probably bought your reader after the new Memory Stick Pro was introduced.
Yes, the new post-Memory-Stick-Pro reader I have can read Memory Stick Pro. The point is I needed to buy new hardware, whereas I didn't have that problem with CompactFlash.
Yeah. First there was Memory Stick. Then came Memory Stick Pro, and my Memory Stick reader couldn't read it, which was a pain as my Dad's camera took Memory Stick Pro.
Then I got a phone with Memory Stick Duo, and guess what? Phone couldn't read Memory Stick or Memory Stick Pro, reader couldn't read Duo.
Then came the Memory Stick Pro Duo.
Meanwhile, my new Canon SLR can use the same CompactFlash cards I was using 4 years ago, and my new 2GB CF card works in my existing card reader.
Really? I kept looking, and never saw any option for buying a ThinkPad without Windows, let alone one with Linux preloaded.
Sure, back in the days of the ThinkPad 600 there were a few abortive experiments, but that was a long time ago.
I think the only difference is that Lenovo have come out and stated what was IBM's unofficial policy for years.
[Opinions mine, not IBMs.]
The thing is, we already have S/MIME. PayPal and the banks could start using it tomorrow if they wanted to. Every Mac user, every Mozilla Thunderbird user, and so on, already has client support.
They don't bother to sign their messages because it doesn't cost them a significant amount if you get scammed.
The possibility of paying massive bandwidth fees to Cingular, for example.
If you're worried about mounting deficits, why not go after a few of the corporations that make billions of dollars in profits, but mysteriously pay no taxes?
For example, News Corporation (Rupert Murdoch, FOX) who pay 0-7% tax. Or Pepsico, who paid 0% tax in 2000, and still pay very little.
Yeah, I've found that when I listen to Sparks, Pandora can't seem to suggest anything else that's remotely appealing. Either Sparks can't be explained by Pandora's categories, or there's nothing else like 'em.
It's better than SLES 9 on the server I've been setting up, in that curses programs like YaST actually work properly without locking up, and the video drivers for Unichrome work.
I always hated YaST for package management, so the option of using APT is a big win. I haven't tried the ZenWorks or 'smart' package managers everyone else is talking about.
Basically, for a server it's OK. I'd rather be running Debian, but the software I need to run is built for SLES.
If the owner wants to write in VB6, or you're not prepared to drink the Microsoft Kool-Aid and require .NET and a 15MB download, you might want to consider RealBASIC.
It's almost totally compatible with VB6, and compiles to native code on Windows, Macintosh and Linux, so you'd actually have the option of supporting multiple platforms.
http://www.realsoftware.com/
Well, I've had two answers to what dnl means, and they're both totally different. So much for self-explanatory config...
Shell scripting isn't hard when it's done appropriately. It becomes a pain when you want to do anything moderately complex, but at that point you should be using a more appropriate tool anyway.
The fact that there are plenty of suckers willing to pay some parasite thousands of dollars for a domain name doesn't change the essentially worthless nature of a domain name, and doesn't change the fact that the people who squat on domains they aren't using in order to try and extort money are lowlifes.
My rule is: Don't feed the parasites. Think of a new name. Remember that none of the biggest most successful Internet companies paid off cybersquatters for "prime virtual real estate" like search.com or books.com; they all went with obscure names nobody had registered like "Google", "ebay" or "Amazon".
In fact, paying big bucks for a domain name is a pretty good predictor of business failure, as far as I can see.
We don't have to imagine such a device, there's one in my living room.
TiVo take Linux and use it to create their platform. They release the code as required, but it's utterly useless, because I can't modify the boot sequence of my TiVo to (say) make it start sshd--it's all locked down with DRM built into the hardware.
I have a perfectly legitimate, legal reason to want to be able to modify the code: I'd like to be able to SSH in to my TiVo from a remote location and tell it to record something.
Thus TiVo have taken GPL code, and prohibited me from modifying it and using it myself on hardware I own. That's exactly what the GPL was supposed to prevent from happening.
[And as for "the market will sort it out" claims--show me the satellite or cable company that provides a DVR that's open and I'll switch. And no, MythTV isn't a sensible option, because that involves decoding the MPEG stream to analog video and then re-encoding it back to MPEG again.]
Except you can, and people do it all the time. Just yesterday I bought a bunch of FLAC files from a music download service, and some CDs. I'm about to order a book. I subscribe to a newspaper online, and a magazine on paper.
DRM isn't about being able to sell intellectual property for a profit, it's about being able to jack the prices up to previously unsustainable levels.
Sendmail is pre-Internet. It was built to route mail between BITNET, UUCP, ARPAnet, JAnet, and so on, all of which had different e-mail syntax. That's why it has a big slow crufty macro engine that every message goes through, and that's why it rewrites the headers of e-mail passing through it. None of that is necessary or desirable these days. Most of sendmail's other problems, from lack of speed to poor security, flow from that initial design decision, so you really need to start again from scratch with a simple e-mail parser and build up from there.
It's still garbage. Sample "improved" sendmail config:
Sample postfix config:
I know which I'd rather edit. I mean, without looking at the manual, I've no idea what that dnl crap is about.
No, the point was to change it just enough to annoy people, but not enough to allow any changes that might fix the basic design issues. Because, you know, it has to look like Slashdot. So the contest was to come up with new lipstick for the pig. When I read the list of restrictions and found out the judges were looking for something pretty much like what they already had, I lost all interest.
That's nothing to do with CSS. Gradients and rounded corners are a bit of a bitch to do in CSS, in fact. They just happen to be the trendy thing in web design at the moment.
Rounded corners are easier with tables than with CSS. (There, I admitted it.)
So don't configure your browser to use them as the default, and Slashdot won't use them.
Then there's Dark Star. Personally, I prefer the longer version there, even though it was padded out to make it long enough for movie theaters.
I think they could have phrased that more tastefully.
And what about the highest performance file systems (XFS, ReiserFS) which don't?
Getting good performance out of clustering and padding is hard. Implementing high performance filesystem-style clustering for the file writing code of an office suite is ridiculously overoptimizing the wrong piece of code.