Ok, I guess you're right. The main thing is that it seemed sci fi writers didn't predict the exponential miniturization and performance curves before they had already started to happen.
Yeah, but these guys were, what, 5 years ahead of the curve? It's not like they invented the helicopter 500 years ago. People like Arthur C Clarke were decades ahead of their time in (seemingly) all areas except computers.
Good work, Nostradamus. I'm going to predict that, between now and June, the world's climate will gradually get warmer in the northern hemisphere and colder in the southern.
If you know your technology can detect planets, say, 100% larger than Earth, and the technology advances by a factor of two every n years, then you can predict that you will find Earth-sized planets in n years. It's not rocket science.
I'm no expert, but I think of it this way: a holographic plate has certain properties that cause it to emit/transmit/reflect light in the same way as if it had come from a real 3D object. The effect is as though the plate were a window through which you look at the object.
Presumably they could also make the object appear in front of the plate, with the appropriate manipulations.
Incidentally, this also explains that curious property of holograms that you can break one in half and get two full holograms. The effect is like looking through two smaller windows instead of one large one: you can still see the whole object behind the window; you just have less parallax.
So, it's hard, being a fairly non-commercial distro to get a group of developers motivated enough to code an autodetect routine that works for everyone. RedHat and Mandrake have done it, because they are trying to meet the needs of their market. The debian people aren't aiming for the same things.
Though, through the magic of Open Source(tm), I suppose Debian could just steal the autodetection code.
So add autodetection to Debian already! What are you waiting for? It's open-source.
Hey, don't change the subject! I'm not demanding autodetection. I'm just saying that having autodetection is better than not having it, at least in the installer (see below).
On the other hand it gets on my nerves that the autodetect routine is going to try to redetect my hardware all the time, so then I'd have to go hunt down the init script and disable it.
Hey, if you're going to use Linux, you should know what's running on your box.:-)
Seriously, I agree with you on this point, and I consider it problematic for the exact same reason as the autodetection thing: I don't want to spend my time tracking down an init script any more than I want to spend my time configuring the hardware.
Anyway, what I had in mind was for the installer to do the autodetection, not the init scripts.
But, if you're really trying to get it to do something, like run Apache or ftpd or whatever, then it's worth the effort, because once you're done, you don't have to mess with it.
Agreed. However, wouldn't it be even better if you didn't have to mess with it in the first place? Actually, Apache is a pretty good example. I wanted to make a website for an Unreal Tournament clan one time. On debian, I just installed the Apache, PHP, and MySQL packages, and the whole lot just worked. It was only when I wanted to get fancy that I needed to learn about the config files.
In contrast, there is no analogous thing for my network card. If I don't configure it (ie. choose the right driver), it doesn't work.
If you are installing on a number of different boxes, then you're probably doing it from work right? If so, and you'll be running debian on those boxes, deal with it. That's your job. If not, then deal with it, or pick a distro that works how you like it to work.
The problem is that Debian doesn't autodetect. Your solution, apparently, is that people who want autodetection shouldn't use Debian. I would humbly submit that adding autodetection to Debian is a better solution.
If these are your home boxes, you should know what's in them or you should find out and make yourself a notebook that contains a list of all of your systems and all of their relevant chipsets (mainboard chipset, processor, network, video, sound, etc).
What the hell are you talking about? I thought computers were supposed to automate things.
Take a good, hard look at what you have written and tell me again, with a straight face, that autodetection is not a good idea.
Do you want control of your system or not?
I'd rather spend my time using my computer than trying to configure it to work properly. I just spend my whole damn weekend doing just that, and if I never have to do it again it will be too soon.
if you don't KNOW what's in your box, why are you installing on it?
Perhaps two scenarios:
You are installing on a number of different boxes, and the labour involved in finding out exactly what is in each box is multiplied, especially if you make a mistake.
You know the manufacturer of a device, but not which driver it should use. (eg. does my D-Link network card use the Tulip or VIA Rhine chipset?)
Yes, modular software is better. However, good luck finding any in the Real World. People think modularity has to be this heaviweight headache like COM or CORBA, or perhaps Java.
I think it is not widely understood that modularity can be fine-grained and pervasive while still being efficient, especially if it is done in such a way that module boundaries can be optimized away (eg. through inlining) at compile time.
(Incidentally, speaking of inlining, gcc is actually a pretty good compiler to use if you want to take this approach. It has great inlining facilities, which makes up for its lackluster optimization capabilities. I had a system I wrote using fine-grained modularity that relied very heavily on inlining for performance. When I moved from gcc to icc (Intel's own C compiler) it became about ten times slower, even after some fairly careful tweaking of icc's command line options. The difference was that icc simply refused to inline most of my function calls for a variety of reasons.)
If you read the article, you'll see that Stallman argues that patents don't encourage innovation. Any thoughts on this?
I admit that I didn't read the article slowly or carefully, but that's not how I understood it.
Well, he said this:
There is economic research showing how imposing the patent system on a field where there is incremental innovation can retard progress. Advocates say: "yes, there might be problems, but the patent system promotes innovation so the problems don't matter."
In fact, there is no reason to believe the patent system does promote progress.
It's in our best collective interest to encourage innovation in software by whatever means we find appropriate, including granting exclusive rights in the form of patents.
If you read the article, you'll see that Stallman argues that patents don't encourage innovation. Any thoughts on this?
Slashdot Mirror
Ok, I guess you're right. The main thing is that it seemed sci fi writers didn't predict the exponential miniturization and performance curves before they had already started to happen.
Yeah, but these guys were, what, 5 years ahead of the curve? It's not like they invented the helicopter 500 years ago. People like Arthur C Clarke were decades ahead of their time in (seemingly) all areas except computers.
Good work, Nostradamus. I'm going to predict that, between now and June, the world's climate will gradually get warmer in the northern hemisphere and colder in the southern.
Good point. Also, Google is, by its nature, always a little out of date.
That's your fault man. Java is the wrong language for this kind of text processing. If you like OO, you should have tried Python.
Just go to Google and add site:slashdot.org to your search.
Mods, I think the parent may be a gem lost in the abyss of anonymity. Do what you think is right.
I doubt it. That's their revenue model. If they give away the engine source, what do they have left?
Thanks for the 21 figures of precision.
What is AIRC? "As I Recall Correctly"?
ERROR: Too many double-negatives. Unable to parse parent post.
If you know your technology can detect planets, say, 100% larger than Earth, and the technology advances by a factor of two every n years, then you can predict that you will find Earth-sized planets in n years. It's not rocket science.
I'm no expert, but I think of it this way: a holographic plate has certain properties that cause it to emit/transmit/reflect light in the same way as if it had come from a real 3D object. The effect is as though the plate were a window through which you look at the object.
Presumably they could also make the object appear in front of the plate, with the appropriate manipulations.
Incidentally, this also explains that curious property of holograms that you can break one in half and get two full holograms. The effect is like looking through two smaller windows instead of one large one: you can still see the whole object behind the window; you just have less parallax.
That's a good trick. I'll keep it in mind. Thanks!
Seriously, I agree with you on this point, and I consider it problematic for the exact same reason as the autodetection thing: I don't want to spend my time tracking down an init script any more than I want to spend my time configuring the hardware.
Anyway, what I had in mind was for the installer to do the autodetection, not the init scripts.
Agreed. However, wouldn't it be even better if you didn't have to mess with it in the first place? Actually, Apache is a pretty good example. I wanted to make a website for an Unreal Tournament clan one time. On debian, I just installed the Apache, PHP, and MySQL packages, and the whole lot just worked. It was only when I wanted to get fancy that I needed to learn about the config files. In contrast, there is no analogous thing for my network card. If I don't configure it (ie. choose the right driver), it doesn't work.True, but I suppose if it did that, it could probably do the autodetection without much additional effort, and all this would be moot.
Take a good, hard look at what you have written and tell me again, with a straight face, that autodetection is not a good idea.
I'd rather spend my time using my computer than trying to configure it to work properly. I just spend my whole damn weekend doing just that, and if I never have to do it again it will be too soon.I just can't stop laughing.
Yes, modular software is better. However, good luck finding any in the Real World. People think modularity has to be this heaviweight headache like COM or CORBA, or perhaps Java.
I think it is not widely understood that modularity can be fine-grained and pervasive while still being efficient, especially if it is done in such a way that module boundaries can be optimized away (eg. through inlining) at compile time.
(Incidentally, speaking of inlining, gcc is actually a pretty good compiler to use if you want to take this approach. It has great inlining facilities, which makes up for its lackluster optimization capabilities. I had a system I wrote using fine-grained modularity that relied very heavily on inlining for performance. When I moved from gcc to icc (Intel's own C compiler) it became about ten times slower, even after some fairly careful tweaking of icc's command line options. The difference was that icc simply refused to inline most of my function calls for a variety of reasons.)