"Nick Rolfe" wrote:

http://morpheus.cs.ucdavis.edu/papers/sweeny.pdf

He refers to Haskell and its strengths (and some of its weaknesses) quite a bit.

For those who don't know him, Tim Sweeney is the main programmer behind Epic Games's popular Unreal Engine. When he talks, many game developers will listen.

We will dream, most likely.

Perhaps more importantly, anything he does will affect a large number of game developers.

Dreaming of pointy-haired bosses listening to him, that is. That said, he made the slides before the advent of the GHC API, which is the dream of anyone being worried about scripting performance... not to mention that it greatly reduces the edit/compile/test cycle once in place. Bytestring fusion wasn't in sight, too, afair... or spj worrying about pipeline stalls, to single out a single thing. All this is quite important in an industry where you test whether it's faster to properly generate the index numbers or rely on Java exceptions when you want to randomly index 3x3 portions of an NxM map where the indexed portion reduces to 2x3 or 2x2 in rare cases. -- (c) this sig last receiving data processing entity. Inspect headers for past copyright information. All rights reserved. Unauthorised copying, hiring, renting, public performance and/or broadcasting of this signature prohibited.

"Sebastian Sylvan" wrote:

...

...

For those who don't know him, Tim Sweeney is the main programmer behind Epic Games's popular Unreal Engine. When he talks, many game developers will listen.

We will dream, most likely.

...

Perhaps more importantly, anything he does will affect a large number of game developers.

Dreaming of pointy-haired bosses listening to him, that is.

I think he meant in the sense that the unreal engine has *lots* of licensees. If UnrealScript 4 is a Haskell-like language with lenient evaluation and limited dependent typing, etc., then that's what a large number of game developers will use in their day-to-day work.

Er, yes. Some gameplay programmers will, and also some level designers, seen 3d-wise, that most likely means gfx guys, who would generally rather work with some warm, fuzzy, graphical switch->event thingy. The rest is still afaict left with C++ which gets linked into the engine. You make less bugs with that language? Fucking learn to write C++! I don't know how the UE is priced, but with the $10.000 that the cheapest id engine cost as I looked it up you can develop a whole game... CrystalSpace is sufficient for nearly everything one could want to do, except maybe ahead-of-the-art graphic demos pushed into markets by million-euro advertisement and game-mag-bribing budgets. I don't think you would get even close to 1000 people who use UnrealScript to earn their living. -- (c) this sig last receiving data processing entity. Inspect headers for past copyright information. All rights reserved. Unauthorised copying, hiring, renting, public performance and/or broadcasting of this signature prohibited.

"Sebastian Sylvan" wrote:

...

You make less bugs with that language? Fucking learn to write C++!

Excuse me?

A probable exclamation of a pointy-haired boss, that is. What I wanted to say is that if you tell such a guy that you'll make less bugs in language X, he would assume that you can't program properly at all or in the language you're supposed to be programming in. As much as I agree with your idealism, the most rigorous proof of the earth being a sphere orbiting the sun could not convince people of the fact that the earth isn't a planar disk, and actual awareness on how the sun "rises" and the sun "sets" is rare even in these days where people accept the fact. Paradigm changes might only take a generation in a scientific community, but it takes close to eons in a society, as its members aren't nearly as involved in the scientific progress, and are (that's a guess now) equally likely to resist a new paradigm as the scientists themselves. The surest thing to make people switch is to make them not aware of it, i.e. make things look exactly like in C, with incremental updates of the same variable and everything, while still retaining a purely functional semantic under the hood. I guess that's why success has to be avoided at all cost. -- (c) this sig last receiving data processing entity. Inspect headers for past copyright information. All rights reserved. Unauthorised copying, hiring, renting, public performance and/or broadcasting of this signature prohibited.

On Jan 10, 2008 1:49 PM, Achim Schneider wrote:

"Sebastian Sylvan" wrote:

...

...

You make less bugs with that language? Fucking learn to write C++!

Excuse me?

A probable exclamation of a pointy-haired boss, that is. What I wanted to say is that if you tell such a guy that you'll make less bugs in language X, he would assume that you can't program properly at all or in the language you're supposed to be programming in.

Maybe I'm just lucky, but if we are still talking about the games industry I don't think this fits my experience of bosses. Games compete very much on performance, and we basically rewrite almost all of our code over a few years or so anyway (though not all at once). I think that means the games industry is very well suited to be at the forefront of adopting new technology, but there would have to be a fairly disruptive change in order for someone to say, for example, "right, let's rewrite all our runtime code in Haskell". Concurrency does seem pretty disruptive. -- Sebastian Sylvan +44(0)7857-300802 UIN: 44640862

barsoap:

"Sebastian Sylvan" wrote:

...

Concurrency does seem pretty disruptive.

Yes, the thought of using par on a dual quad-core makes me salivate.

Haskell is (in a very small part) driving sales of multicore boxes -- I've met half a dozen people who nominated Haskell's multicore support as a deciding reason to explicit get a multi-core box. Maybe we need to get some kickbacks from the hardware vendors, for helping their sales :) Intel, AMD, Nvidia -- are you guys listening?? In return, we can improve the parallelism support further. --Don

On Jan 10, 2008, at 1:57 PM, Achim Schneider wrote:

Now don't make me think of using par on a beowolf cluster of ps3's.

Never in my life have I _literally_ drooled over using a programming abstraction. - Jake

Ketil Malde wrote:

"Sebastian Sylvan" writes:

...

Maybe I'm just lucky, but if we are still talking about the games industry I don't think this fits my experience of bosses. Games compete very much on performance, and we basically rewrite almost all of our code over a few years or so anyway

Another thing is that having the next big game means an incredible amount of money - more than a Hollywood blockbuster, according to popular rumor. Many companies may be willing to take a risk on promising but immature technology in the hope that it will give them the advantage they need to deliver the next Halo (or whatever).

Very competitive industry, large money involved, lots of software rewritten or developed from scratch. Sounds ideal. Any boss who insists on doing things the way they've always done it won't last long.

Just to set things straight: You can make more money selling mind trainers and party games like Nintendo does, or selling games with slideshow complexity but much visible flesh for handsets... at least here in Germany that's the much bigger portion of the industry. -- (c) this sig last receiving data processing entity. Inspect headers for past copyright information. All rights reserved. Unauthorised copying, hiring, renting, public performance and/or broadcasting of this signature prohibited.