It outlines how Largo, in Florida, uses KDE 2.1.1 on 400 NCD thin clients to cater for 800 users.
Not to mention this/. thread. Unfortunately from the open-source perspective, they're using these Linux terminals primarily to run commercial software packages: WordPerfect, Excel, PowerPoint, and Access.
Thanks for the info. I checked the WordPerfect system requirements and it said that Windows is the only supported operating system. However, I now see on the full products listing page that there is a Linux version, which is not referenced from the main WordPerfect page.
The Linux version of WordPerfect appears to be an end of line product. The Windows versions have been updated while the Linux version has not, and the product is not for sale in Corel's online store. In any case, it is commercial, not open source.
I had a hard time finding out exactly what application software was being used based on the article and its supporting materials. However, when I saw references to Citrix and Windows terminal server, I could see that this was not the free-software, Microsoft-free paradise one would expect from the glowing summary here. As you've confirmed, the users are actually running Microsoft Office and other commercial software in a remote mode. I apologize for getting the Windows software in use wrong, but I think my main point still stands. Even if regular users are able to get by with KDE as a desktop shell these days -- and I don't doubt it -- when it comes to the applications they need to do their work, they're still dependent on commercial software, and on Microsoft.
I was surprised to read through this thread at 3 and find not a single reference to the fact that this system is serving up Windows software (WordPerfect) to Linux thin clients. It's hardly a Microsoft-free utopia. In fact it's primarily Windows software running remotely. So I'm not sure what point this is supposed to make about the prospects of Linux as a desktop operating system.
As long as your hardware is not total crap, Linux is not far from being a "insert CD, click install, come back in 15 minutes, start working" kinda OS. What do I mean by crap hardware? The kind of stuff that no self-respecting kernel hacker would buy, let alone write drivers for. (Like Winmodems, no-name Ethernet cards, old cheap SB-compatible sound cards, scanners with proprietary interfaces made by some company that died 5 years ago, etc.)
This is an internal myth of the Linux community. There is nothing wrong with reducing the cost of modem hardware by offloading some of its functions onto the main processor. In fact there's a major user benefit, which is lower cost. The reason people say Winmodems are crap is so they don't have to deal with the issue that Linux software support and availability isn't as good as on Windows. The tiny Linux market share doesn't lend itself to broad software or hardware support.
We have the same problem on the Mac side of the fence. It's a really unfair thing in a lot of ways, but it also is a concrete problem with using a minority platform, and the way to deal with it is not by saying that all the missing hardware and software is crap. (Although that was a good enough answer back when the war was between Mac and DOS!)
Mozilla, btw, is moving along nicely.
An ever-increasing number of bugs is not my idea of moving along nicely. I wouldn't ship commercial software that had Mozilla's defect curve. I'd link directly to the chart, but the bug chart feature is broken again today.
I think it's time to face up to the fact that projects people do in their spare time as tinkerers may never catch up to those that are funded, staffed and managed based on the potential for financial reward.
You need to read Eric Raymond's Cathedral and the Bazaar. http://tuxedo.org/~esr/ Enjoy.
I read it years ago. It bears no resemblance to reality and has even largely fallen out of favor in the open source community.
His understanding of what it means to manage a project is totally wrong.
His example program is a trivial small utility program having nothing to do with large software projects.
His dictum that "to many eyes, all bugs are shallow" is demonstrably false.
His predictions of commercial benefits from open source have caused several companies to crash and burn, but none to achieve profitability.
His approach never had a quantifiable business model.
He is completely unaware of software engineering as a discipline.
He acknowledges that he has never taken a single class in the subject, and it really shows.
I asked why there should have been a controversy about whether the Linux community existed. It seems a community is a tangible thing. No one bothered to respond, but I continued to ask myself the question, and realized that there was an answer. The attendance of nearly two thousand people at the 2001 O'Reilly Open Source Convention demonstrates the existence of a community, though not necessarily a large one.
Also, consider this: I get paid $100/hour for consulting. What does it cost me to muck around with my X configuration for a few hours to get it to work with my video card?... A few hours of not having to screw with that stuff, and the Apple hardware suddenly isn't a premium investment after all. In my business, I really have to consider the Total Cost of Ownership.
Exactly. I keep trying to explain this here -- in fact, I've been trying to explain it to people for about ten years. Back in the early 1990's, people said Macs were sooo expensive and slow compared to PCs. I moved from a Mac to a Windows shop, and at the time, I noticed that everyone on PCs was spending literally hours a day doing nothing but tweaking their configurations to keep them running. Every time there was any software install, that person would lose at least a day of work; add a peripheral, and the impact was a week. It wasn't even controversial -- the PC architect took those as the stated costs of using Windows, and I'm just quoting his rules.
By contrast, in the Mac shops I was used to, the Macs just worked, and the Mac users could just do their work. Installing software and peripherals had costs measured in minutes instead of days. "Now how much would you pay?" What do you think the productivity cost per user is in that environment? Linux today seems to be barely above that early-90's Windows 3.x TCO level.
Despite the fact that the GNU project has been around for every bit as long as the Mac (since 1984), they still don't have an OS that my mom could install and use. Why is that?
We keep getting told that free software improves so rapidly, but somehow its user experience still hasn't reached the level of a commercial operating system from seventeen years ago, and its compiler hasn't reached the performance (compilation or runtime) of commercial compilers ten years ago. I think it's time to face up to the fact that projects people do in their spare time as tinkerers may never catch up to those that are funded, staffed and managed based on the potential for financial reward.
(This leaves open the question of free software that is funded in hopes it will provide commercial benefit, but looking at Mozilla as an exemplar would not result in a positive conclusion.)
Responding to myself in an old thread, but I wanted to add a data point.
Thinking this over, I realizes that no one would ask whether there was a Mac community because you can just go to MacWorld and see, touch, hear, smell and photograph it. So couldn't you do the same with the Linux community?
(Well, maybe you wouldn't want to do the smell part, but you know what I mean.)
So I asked a friendly O'Reilly representative how many people registered at their 2001 Open Source Convention. The answer, including attendees, speakers, press, exhibitors and expo only, was 1800. That's small compared to, say, attendees of MacWorld or the World Science Fiction Convention, but it's a respectable number. It also seems reasonable to conflate Linux and Open Source for this purpose.
Given this, it does not require reference to St. Anselm or the Primum Mobile to verify that the Linux community exists. That someone would prefer a religious argument to an empirical one in this context is interesting, but further deponent sayeth not.
Maybe he did it in his spare time.
Maybe he did it for a sense of accomplishment.
Maybe so, but we were discussing benefits from the ability to modify open source. What you're describing is the benefit of having a hobby, not an economic benefit. We all know that some people enjoy tinkering with code in their free hours. Presumably, we also know that most people would rather have sex, read a book or watch a movie. So the benefit of this opportunity doesn't extend beyond a tiny subculture.
Maybe he did it and contributed it back to the keepers of the source, they implemented it, thus saving hundreds if not thousands of other people to have to go out and buy a new sound card.
That could be, and if so it would be an altruistic act, conferring a benefit through the economy of replication. Did he, though? It seems all he wanted was to get sound on his own system.
There is a theoretical benefit to society from code sharing but it doesn't seem to have fulfilled the promise of systems with high reliability and low TCO yet. Part of the reason may be related to all the complaints from people who can't get their changes merged into the trunk of their favorite project. It seems to have become such a hassle that many people have stopped trying.
(And BTW, why the heck would someone have to read source code to understand how a kernel call works? Shouldn't he have been able to settle this using kernel docs rather than reading the source?)
I tracked down where this could happen in the driver, and then tracked the corresponding function in the Kernel source. Turned out to be an error in the driver, a one-liner. Fixed it, recompiled and bingo - sound in Linux.
Um, d00d, one hour of C programmer time costs a lot more than a new sound card. What did you save again? Negative a hundred dollars?
If it was an open source program, it would be possible to hire a programmer to add those features....
I'd like to see you try this experiment sometime. Get back to us after the $20,000 and one month you expected to spend has ballooned into six months and $150,000, without delivering you any features in a form you want to use.
We have a new ability for real-time signal handlers in user tasks, where the user process makes an rtlinux_sigaction which will work like the POSIX sigaction, except the signal handlers run in realtime and there's an extension to allow us to catch periodic interrupts. As a result, the user process can designate a function to operate within hard real-time deadlines. And those functions run in the address space of that process, so they can share data with the process, and call functions libraries of the process.
That'll be very useful for high-bandwidth multimedia playback, which currently seems to be a problem for some UNIX-based systems such as Mac OS X. Is anyone looking at a Darwin port?
The most interesting part of the article for me was this:
I did respond to Reichard after he demanded proof that the Linux community existed. Proof, that is, after saying more or less that he hated to be the one to say there was no Santa, but there was no Linux community.
I ended up saying that if he chose not to believe in the community that was his prerogative. Either we were serving our readers or we weren't. As for me convincing him of the fact, I compared proving that the community existed to proving that God existed.
That seems very curious. I don't believe in God, because there's no evidence of its existence. God is some far-off transcendental entity about whom no reliable information can be gathered. But a community here on earth is a different matter. It leaves traces, it walks around; it can be photographed. Why would demonstrating its existence be anything like proving the existence of God?
I'm not sure how to take this, exactly. Is Paul Ferris just bad at documenting the evidence of his senses? Why would it be so hard for people professionally engaged with the "Linux community" to even muster evidence that the community exists? If there isn't something better than a religious assertion to fall back on, then it seems reasonable to say that it does not exist.
One piece of evidence is the source code for the kernel. That seems to be the work of less than a thousand people, spread out all over the world. That would be a tiny community by anyone's standards, but it does seem to exist. If it's just a few people on a mailing list, though, it may stretch the term "community" to the breaking point. There's no question that people contribute software to Linux, but is there a community of any significant size? What's the evidence? I ask this in all sincerity.
If you use the installer, you can install over an old install -- the installer deletes the old files.
It's when you install by unpacking a zip archive that you have to make sure you use a lean directory... please get your facts straight before commenting.
Sorry, you are mistaken. From the release notes: Install into a new empty directory. Installing on top of previously installed builds may cause problems.
Of course, to prevent breakins, just have the recieving device set up with authentication. Just like you would running a dial-up server at home.
The difference, of course, is that it's harder to tap a phone line for data than to drive around with a radio receiver. For dial-up authentication, passwords are usually sent unencrypted. That would be a disaster for a wireless network.
I believe I also read that these modems encrypt data locally, and transmit on a given frequency for only a very short burst at a time.
The frequency-hoppping element doesn't seem too relevant -- since the modems can establish a connection between each other in an untrusted mode, the same algorithm can be used to intercept the transmission.
If you've got concerns about eavesdropping, then add your own encryption before data is sent.
Data such as the password? At what point would encryption kick in to protect the password negotiation? Any idea what software would be used on both ends for that scenario?
Cool link, but it seems to me that only discusses the on-network mode. What about the peer-to-peer mode? What safety do you have against others initiating connections to your device or MITMing connections you make to others?
Alpha is supposed to mean feature complete software. This software is missing the upgrade-in-place feature, so should we assume it's pre-alpha? It's being rolled out as a late beta, which is supposed to mean that for all intents and purposes it is the final software except for outstanding bug fixes. Assuming that new features will be added in late beta is not a reasonable assumption.
I said it takes money to do good UI design, that it can't be done by lone programmers in a vacuum, and that it takes observation of real users. You respond that that's nonsense, because Apple and Microsoft had to spend a lot of money, techies can't do it themselves, and it comes down to the real users.
As far as I can tell, we're agreeing, but your message is somehow phrased as a flame against mine.....
If you know of some programs you think are hard to use, fix them!
Great! Should I tell the user testing labs to bill the facilities, time, gratuities, food and tapes to you?
People don't seem to get the fact that user interface design isn't something you do at home in your underwear. It requires real-world data. There's no way to get that data without spending money.
I always install new releases into their own folder.... This has been recommended practice from day 1, AFAIK.
Recommended in the release notes, perhaps, but it's bad software practice. It creates an unreasonable burden on the user, especially given that the 80% use case for running an installer is a reinstall or upgrade. This practice is acceptable only in the open source world -- no mainstream commercial software vendor could get away with it.
And no, RTFM is not an answer here, any more than it is anywhere else. Software is supposed to be reliable and self-documenting.
This thread reminds me why I don't read science fiction any more. I recall Vinge as a terrible writer, with hackneyed adventure plots, characters with the dimensionality of a cardboard cutout, and themes that reduce to "ain't science grand?" I read through this entire thread at mod level one, and sure enough, there is not a single comment about his stories or his characters, because those barely exist. Apparently no one here minds that little fact. Critical standards in science fiction are abysmally low; the fans routinely gargle boatloads of watery spooge as if it were fine champagne.
It's sad that there's not a better venue for scientific speculation per se. If there were, people with no ear for fiction, such as Vernor Vinge, Robert Forward and Isaac Asimov, would not feel themselves forced into quasi-fictional exercises that demean both themselves and the storyteller's art.
The brain is also a finite machine - we will soon be able to build electronics that excess the capacity.
Which is completely irrelevant. The human nervous system evolved over the course of billions of years and works in a very specific and detailed fashion, most of which is still a mystery to us at the computational level. Without reproducing all that evolved design, we would not have anything like human intelligence.
We already have machines that exceed the human capacity in every way physically, but we have not yet been able to create a robotic construction worker. Why should just throwing terraflops at the intelligence problem go any further towards solving it?
Forget the grotesquely inflated values of bubble IPOs and go back just ten years. IPO prices for software companies were routinely running in the tens of millions then, and the desktop and server markets have grown enormously in the last ten years.
And here we have an IPO for under 5 million Euros. That's not even in the ballpark for what a software company would have IPO'd for ten years ago.
It's pretty good for a company that has always lost money, I suppose, but it doesn't exactly hold out the prospect of a brilliant future.
Slashdot Mirror
Not to mention this /. thread. Unfortunately from the open-source perspective, they're using these Linux terminals primarily to run commercial software packages: WordPerfect, Excel, PowerPoint, and Access.
Tim
If you screen shot the graph and draw out the lines at the current slope, you'll see that they meet in mid-November.
And we all know what happens when the beams cross.
Tim
Any idea whether this will have an effect on Mozilla or the Open Directory Project?
Tim
The Linux version of WordPerfect appears to be an end of line product. The Windows versions have been updated while the Linux version has not, and the product is not for sale in Corel's online store. In any case, it is commercial, not open source.
I had a hard time finding out exactly what application software was being used based on the article and its supporting materials. However, when I saw references to Citrix and Windows terminal server, I could see that this was not the free-software, Microsoft-free paradise one would expect from the glowing summary here. As you've confirmed, the users are actually running Microsoft Office and other commercial software in a remote mode. I apologize for getting the Windows software in use wrong, but I think my main point still stands. Even if regular users are able to get by with KDE as a desktop shell these days -- and I don't doubt it -- when it comes to the applications they need to do their work, they're still dependent on commercial software, and on Microsoft.
Tim
Tim
Tim
No, although they certainly are improving. You might want to read the GNOME Usability Study Report.
As long as your hardware is not total crap, Linux is not far from being a "insert CD, click install, come back in 15 minutes, start working" kinda OS. What do I mean by crap hardware? The kind of stuff that no self-respecting kernel hacker would buy, let alone write drivers for. (Like Winmodems, no-name Ethernet cards, old cheap SB-compatible sound cards, scanners with proprietary interfaces made by some company that died 5 years ago, etc.)
This is an internal myth of the Linux community. There is nothing wrong with reducing the cost of modem hardware by offloading some of its functions onto the main processor. In fact there's a major user benefit, which is lower cost. The reason people say Winmodems are crap is so they don't have to deal with the issue that Linux software support and availability isn't as good as on Windows. The tiny Linux market share doesn't lend itself to broad software or hardware support.
We have the same problem on the Mac side of the fence. It's a really unfair thing in a lot of ways, but it also is a concrete problem with using a minority platform, and the way to deal with it is not by saying that all the missing hardware and software is crap. (Although that was a good enough answer back when the war was between Mac and DOS!)
Mozilla, btw, is moving along nicely.
An ever-increasing number of bugs is not my idea of moving along nicely. I wouldn't ship commercial software that had Mozilla's defect curve. I'd link directly to the chart, but the bug chart feature is broken again today.
I think it's time to face up to the fact that projects people do in their spare time as tinkerers may never catch up to those that are funded, staffed and managed based on the potential for financial reward.
You need to read Eric Raymond's Cathedral and the Bazaar. http://tuxedo.org/~esr/ Enjoy.
I read it years ago. It bears no resemblance to reality and has even largely fallen out of favor in the open source community.
Tim
Tim
Exactly. I keep trying to explain this here -- in fact, I've been trying to explain it to people for about ten years. Back in the early 1990's, people said Macs were sooo expensive and slow compared to PCs. I moved from a Mac to a Windows shop, and at the time, I noticed that everyone on PCs was spending literally hours a day doing nothing but tweaking their configurations to keep them running. Every time there was any software install, that person would lose at least a day of work; add a peripheral, and the impact was a week. It wasn't even controversial -- the PC architect took those as the stated costs of using Windows, and I'm just quoting his rules.
By contrast, in the Mac shops I was used to, the Macs just worked, and the Mac users could just do their work. Installing software and peripherals had costs measured in minutes instead of days. "Now how much would you pay?" What do you think the productivity cost per user is in that environment? Linux today seems to be barely above that early-90's Windows 3.x TCO level.
Despite the fact that the GNU project has been around for every bit as long as the Mac (since 1984), they still don't have an OS that my mom could install and use. Why is that?
We keep getting told that free software improves so rapidly, but somehow its user experience still hasn't reached the level of a commercial operating system from seventeen years ago, and its compiler hasn't reached the performance (compilation or runtime) of commercial compilers ten years ago. I think it's time to face up to the fact that projects people do in their spare time as tinkerers may never catch up to those that are funded, staffed and managed based on the potential for financial reward.
(This leaves open the question of free software that is funded in hopes it will provide commercial benefit, but looking at Mozilla as an exemplar would not result in a positive conclusion.)
Tim
Thinking this over, I realizes that no one would ask whether there was a Mac community because you can just go to MacWorld and see, touch, hear, smell and photograph it. So couldn't you do the same with the Linux community?
(Well, maybe you wouldn't want to do the smell part, but you know what I mean.)
So I asked a friendly O'Reilly representative how many people registered at their 2001 Open Source Convention. The answer, including attendees, speakers, press, exhibitors and expo only, was 1800. That's small compared to, say, attendees of MacWorld or the World Science Fiction Convention, but it's a respectable number. It also seems reasonable to conflate Linux and Open Source for this purpose.
Given this, it does not require reference to St. Anselm or the Primum Mobile to verify that the Linux community exists. That someone would prefer a religious argument to an empirical one in this context is interesting, but further deponent sayeth not.
Tim
Maybe so, but we were discussing benefits from the ability to modify open source. What you're describing is the benefit of having a hobby, not an economic benefit. We all know that some people enjoy tinkering with code in their free hours. Presumably, we also know that most people would rather have sex, read a book or watch a movie. So the benefit of this opportunity doesn't extend beyond a tiny subculture.
Maybe he did it and contributed it back to the keepers of the source, they implemented it, thus saving hundreds if not thousands of other people to have to go out and buy a new sound card.
That could be, and if so it would be an altruistic act, conferring a benefit through the economy of replication. Did he, though? It seems all he wanted was to get sound on his own system.
There is a theoretical benefit to society from code sharing but it doesn't seem to have fulfilled the promise of systems with high reliability and low TCO yet. Part of the reason may be related to all the complaints from people who can't get their changes merged into the trunk of their favorite project. It seems to have become such a hassle that many people have stopped trying.
(And BTW, why the heck would someone have to read source code to understand how a kernel call works? Shouldn't he have been able to settle this using kernel docs rather than reading the source?)
Tim
Um, d00d, one hour of C programmer time costs a lot more than a new sound card. What did you save again? Negative a hundred dollars?
Tim
I'd like to see you try this experiment sometime. Get back to us after the $20,000 and one month you expected to spend has ballooned into six months and $150,000, without delivering you any features in a form you want to use.
Tim
That'll be very useful for high-bandwidth multimedia playback, which currently seems to be a problem for some UNIX-based systems such as Mac OS X. Is anyone looking at a Darwin port?
Tim
That seems very curious. I don't believe in God, because there's no evidence of its existence. God is some far-off transcendental entity about whom no reliable information can be gathered. But a community here on earth is a different matter. It leaves traces, it walks around; it can be photographed. Why would demonstrating its existence be anything like proving the existence of God?
I'm not sure how to take this, exactly. Is Paul Ferris just bad at documenting the evidence of his senses? Why would it be so hard for people professionally engaged with the "Linux community" to even muster evidence that the community exists? If there isn't something better than a religious assertion to fall back on, then it seems reasonable to say that it does not exist.
One piece of evidence is the source code for the kernel. That seems to be the work of less than a thousand people, spread out all over the world. That would be a tiny community by anyone's standards, but it does seem to exist. If it's just a few people on a mailing list, though, it may stretch the term "community" to the breaking point. There's no question that people contribute software to Linux, but is there a community of any significant size? What's the evidence? I ask this in all sincerity.
Tim
Sorry, you are mistaken. From the release notes: Install into a new empty directory. Installing on top of previously installed builds may cause problems.
Tim
The difference, of course, is that it's harder to tap a phone line for data than to drive around with a radio receiver. For dial-up authentication, passwords are usually sent unencrypted. That would be a disaster for a wireless network.
I believe I also read that these modems encrypt data locally, and transmit on a given frequency for only a very short burst at a time.
The frequency-hoppping element doesn't seem too relevant -- since the modems can establish a connection between each other in an untrusted mode, the same algorithm can be used to intercept the transmission.
If you've got concerns about eavesdropping, then add your own encryption before data is sent.
Data such as the password? At what point would encryption kick in to protect the password negotiation? Any idea what software would be used on both ends for that scenario?
Tim
Tim
Tim
I said it takes money to do good UI design, that it can't be done by lone programmers in a vacuum, and that it takes observation of real users. You respond that that's nonsense, because Apple and Microsoft had to spend a lot of money, techies can't do it themselves, and it comes down to the real users.
As far as I can tell, we're agreeing, but your message is somehow phrased as a flame against mine.....
Tim
Great! Should I tell the user testing labs to bill the facilities, time, gratuities, food and tapes to you?
People don't seem to get the fact that user interface design isn't something you do at home in your underwear. It requires real-world data. There's no way to get that data without spending money.
Tim
Recommended in the release notes, perhaps, but it's bad software practice. It creates an unreasonable burden on the user, especially given that the 80% use case for running an installer is a reinstall or upgrade. This practice is acceptable only in the open source world -- no mainstream commercial software vendor could get away with it.
And no, RTFM is not an answer here, any more than it is anywhere else. Software is supposed to be reliable and self-documenting.
Tim
It's sad that there's not a better venue for scientific speculation per se. If there were, people with no ear for fiction, such as Vernor Vinge, Robert Forward and Isaac Asimov, would not feel themselves forced into quasi-fictional exercises that demean both themselves and the storyteller's art.
Tim
Which is completely irrelevant. The human nervous system evolved over the course of billions of years and works in a very specific and detailed fashion, most of which is still a mystery to us at the computational level. Without reproducing all that evolved design, we would not have anything like human intelligence.
We already have machines that exceed the human capacity in every way physically, but we have not yet been able to create a robotic construction worker. Why should just throwing terraflops at the intelligence problem go any further towards solving it?
Tim
And here we have an IPO for under 5 million Euros. That's not even in the ballpark for what a software company would have IPO'd for ten years ago.
It's pretty good for a company that has always lost money, I suppose, but it doesn't exactly hold out the prospect of a brilliant future.
Tim