Forgot your password?
typodupeerror

Comment: Re:Acorn Risc Machine (Score 1) 111

by Eythian (#46329221) Attached to: The Ever So Unlikely Tale of How ARM Came To Rule the World

You could jump in and start writing applications on RiscOS directly in BASIC if you wanted. I dabbled a bit with it, but didn't have any access to documentation, so only got so far from reverse-engineering (by which I mean reading the source once you figure out shift-double-click shows you inside the !Application bundle.)

It was a very well done OS though. In some ways, it feels like systems now are only just starting to catch up, and in other ways are still a fair bit behind.

Comment: Re:Support (Score 1) 401

by Eythian (#46207081) Attached to: Leonard Nimoy: Smoking Is Illogical

He's not smoking. There's no smoke. There's no tobacco involved at all. That's where the health issues come from.

Nicotine is about as bad for you as caffeine in coffee (which are both bad for you when you have too much, but otherwise not very harmful.)

And on that note, you did notice the bit where he said he's using less over time anyway, right?

Comment: Re:Hard AI (Score 1) 172

by Eythian (#45935393) Attached to: Regex Golf, xkcd, and Peter Norvig

I gave a definition for AI already. In fact, you quoted it. It's not the most precise one, but I don't have to provide a precise one. All I was doing was demonstrating that your implied definition was far, far too narrow.

There are many things that are considered AI that don't match the range you suggested for it. Genetic algorithms, neural networks, language learning systems, robots that take information from their environment, discard it, and drive into walls. They're never* going to think, but they are rightfully AI.

It's not a requirement of a field that it has tight definitions, especially one where different methods that seem in a similar spirit pop up somewhat regularly.

In essence, writing a computer program that uses heuristical and adaptive methods to generate regexes to match/not match specific lists could be a form of genetic programming (though there are other options) which is comfortably under the AI umbrella.

* never say never, but with the scale they're on at the moment, it seems a long ways off.

Comment: Re:Hard AI (Score 2) 172

by Eythian (#45934739) Attached to: Regex Golf, xkcd, and Peter Norvig

That's not true. The field hasn't been stagnant at all. The problem is you're missing most of the field in your assumption of what it is.

AI is not all about making "computer programs that can think" (a.k.a "Strong AI"), it's more about creating systems that can adapt to their environment in order to improve their chance of success. If their environment is getting lists of names and they adjust, without the direct input of a person, to better match them according to some scoring system, that is AI.

It can use statistical methods, evolutionary characteristics, or just adjusting variables according to how wrong they currently are in order to try to match some problem. I'm not even touching symbolic AI here, because I don't understand it nearly so well.

Strong AI is part of the field sure, but certainly not all of it as you imply. It is a very large field.

Comment: Re:Cybrarian (Score 1) 26

by Eythian (#45721297) Attached to: NZ Developers Win 'Koha' Trademark Case

> It's not as robust as normal commercial library systems software

There are two things wrong here :)

a) it is a commercial library system. I get paid to work on it, to host it, to support it. So do many other people and companies. That's about as commercial as you can get.

b) libraries tend to like it more than their previous proprietary system because it is more robust. It doesn't crash (unless you overload it, but it handles that better than many other systems), it doesn't lose branches for days at a time for no good reason, doing repeated Z39.50 queries against it doesn't cause it to die *cough*voyager*cough*. In addition, it looks nice and is nice to use. Have you seen the public catalogues on proprietary systems? They are almost always horrifically ugly, and do things like have sessions in the URL, so you can't send links to someone else, or use multiple tabs reliably.

Koha is not "good enough because you're poor", although it fits that niche too. It's just good because it has more developers, more libraries involved with its development, and is not marketing-driven. Hell, it was totally web-based in 2000, most other ILS software isn't even now.

> Liblime Fauxha.

I'm stealing that. I hope you don't mind :)

Comment: Re:Does not matter (Score 1) 26

by Eythian (#45710991) Attached to: NZ Developers Win 'Koha' Trademark Case

That makes no difference. The claim wasn't over it being a common language word (which it is here), it's over it being deceptive in the trademark sense. i.e. that there is something in the same space with the same name that's well known.

GP's question is perfectly relevant.

Source: I was physically at the Koha trademark hearing.

Nondeterminism means never having to say you are wrong.

Working...