Don't upgrade to 1.2b, however, because downloading is broken under windows 2000 (how the hell did that get through) and if you try to revert 1.1 you'll find the mail client no longer works.
1.1 Seems the best compromise. It's crashier than 1.0.1, but had some can't-live-without mail bug fixes.
I worry about Mozilla's release schedule, however. They seem more concerned with pumping out the releases than bugfixing. Why should I be prompted to download and help test 1.2b when 1.1 is still crashing daily on me?
The disadvantage of standardizing on gcc is that it has been evolving much more quickly over the last five years than most other C++ compilers. The gcc authors are less constrained in terms of not breaking existing codebases using their compiler, compared to, say, MSVC. The code bases I've been involved with have broken more regularly under new gcc upgrades than most other compilers I can recall. (Partly because there have been more gcc upgrades, it's true.)
Now, this isn't all bad. It means gcc tracks the C++ standard more closely than many compilers. It means the gcc team have finally addressed things like the lack of precompiled header support. But it does mean that it's not enough to standardize on gcc -- you have to guess which release will be a good one, and stick with it for a while. This can be hard to judge.
To be honest, at this stage I'm fairly tired with upgrading code to meet the current version of standards compliance under different compilers. It's kind of satisfying the first few times, but that doesn't last. I prefer to work on functionality.
This is a perfectly valid/interesting post, given the slashdot community's focus on patent rights and renderering technology. Why the fuck is it moderated troll?
I mean, I'm getting the impression most posters are just here to go, "ooh, look at the pretty fishies."
Perhaps it's more to fend off the competition. Also, bear in mind that they make Renderman, which is used by other companies.
Pixar is a pretty nasty company when it comes to patenting graphics techniques that are either obvious or that have been previously presented at academic talks or conferences. They're pretty quick to use such patents, too.
It's probably best to steer clear of any technique they're using. (Subdivision surfaces etc.)
> I find that the best program is almost always the shortest.
I disagree.
If you are working on a large code-base, in conjunction with many other people, this is untrue. Over-abstracting or over-generalizing code has the effect of making that code much harder to parse, and much harder to maintain.
Code abstraction is akin to compression. You make it more difficult for people to decrypt (decompress) what the code is doing, unless the abstraction you're using is a common one that everyone can be expected to know. You are also making the code much less flexible. If code is heavily abstracted, any given required modification is more likely to render the abstraction invalid and require a complete rewrite of the code.
This is the real reason Lisp/Scheme/SML/whatever the current flavour of the month is has never caught on in industry in a large way. PL people like these languages because they make it much easier to produce short, dense, powerful code, and this appeals to the mathematician within. There needs to be more human factors research in PL. Or perhaps they should just change the name of the research area. Much PL work is theoretically valuable, even if it is at best orthogonal to producing a good programming language.
Abstraction is a fine tool when applied carefully, but too many people in academia, especially those who have never written a program over one thousand lines, misjudge where the balance point is.
> Then I get to spend the rest of my time posting on >/. on how wrong you clueless Linux users are about > the state of the Windows world.... and rebooting machines.
This guy is a professor? I really hope this article isn't representative of his academic work. I know economics is the dismal science, but really:
> The one big piece of evidence that I didn't have > when we talked before was a half-year year 2002 > number [that appears to indicate a 9.8 percent > decrease in album sales.] There has to be a > caveat in here, which is that I don't know > if this number is correct. It's a half-year > number that I saw in USA Today, from SoundScan.
So his entire paper is based on some number he read in his morning news. Umm. And then:
> If [we] assume that half the computer owners have > CD burners -- a number that I've seen -- you just > double the decline that's already occurred.
I don't know where to start with this. I've read better logic on crop circle web pages.
Where's the science?
Hmm. A quick web search reveals he spends most of his time defending Microsoft.
Heard it a hundred times. It was relevant five years ago, but not today.
Repeat after me: the price/performance curve for commodity graphics hardware is outstripping that for general CPUs. Graphics hardware's "Moore's Law" is winning.
Now, never make the mistake of assuming that an exponential curve will increase indefinitely -- they're all S-curves in the end -- but while it does, movie and PC graphics quality are not both increasing at the same rate. So assuming that the gap between them isn't going to at the very least close a bit is wishful thinking.
With Pixar patenting everything they can lay their hands on, including obvious methods of how you'd use subdivision surfaces in production, and deep shadow maps, the future for rendering R&D looks a little grim unless you're a big company.
Pixar has successfully prevented ExLuna from selling their core rendering product, a cutting-edge renderer based on the renderman standard, and also made them yank BMRT, by suing them for patent infringement, and threatening them with all sorts of other nasty legal stuff. (ExLuna is/was a company run by Larry Gritz, amongst other ex-Pixarites, and Larry was the author of BMRT.)
You won't see BMRT again.
NVIDIA has since bought ExLuna for their personnel and expertise.
Yay Pixar. It's amazing how far they've taken that bogus distributed sampling patent.
Re:Very nice World Cup, but..
on
World Cup Final
·
· Score: 1
Of course, soon the attacking side will nick another trick from rugby; take the free kick as quickly as possible, so the defenders don't have a chance to get back 10. =)
No thanks. I used to watch hockey (go Pens), but have given up on the game as of this season. With the constant clutching and grabbing and fouling, it must be the dullest sport in the world. (Okay, I exaggerate -- there's always golf.) The only winning strategy is interfering with or screening the goalie while someone shoots from outside. Yawn.
I should make it clear I'm talking about the North American version. The larger-rinked European game is probably more fun.
> Pixar tried to make high-speed hardware for years, and we always found it to be a losing game. I > wrote microcode for one of these beasts, a parallel bitslice engine that inspired today's MMX > instructions. We could not keep up with the development of vanilla CPUs, and the CPUs ended > up being more cost effective.
It is indeed hard keeping up with anything in the consumer space, due to economies of scale.
On the other hand, graphics hardware is now squarely in that space. Many of the points you make are very reasonable (e.g. general purpose vs. specialized for real-time), but I wonder if the consumer-oriented nature of current graphics hardware might not render them moot.
My PhD advisor, Paul Heckbert, told tales of writing microcode for the same beasts; it was obviously a scarring experience =). (Interestingly, Paul is currently working for NVIDIA.)
Slashdot Mirror
I think you mean s/unix/emacs/g
All the emacs users I know suffer from RSI. Hmmm =)
A.
Don't upgrade to 1.2b, however, because downloading is broken under windows 2000 (how the hell did that get through) and if you try to revert 1.1 you'll find the mail client no longer works.
1.1 Seems the best compromise. It's crashier than 1.0.1, but had some can't-live-without mail bug fixes.
I worry about Mozilla's release schedule, however. They seem more concerned with pumping out the releases than bugfixing. Why should I be prompted to download and help test 1.2b when 1.1 is still crashing daily on me?
A.
The disadvantage of standardizing on gcc is that it has been evolving much more quickly over the last five years than most other C++ compilers. The gcc authors are less constrained in terms of not breaking existing codebases using their compiler, compared to, say, MSVC. The code bases I've been involved with have broken more regularly under new gcc upgrades than most other compilers I can recall. (Partly because there have been more gcc upgrades, it's true.)
Now, this isn't all bad. It means gcc tracks the C++ standard more closely than many compilers. It means the gcc team have finally addressed things like the lack of precompiled header support. But it does mean that it's not enough to standardize on gcc -- you have to guess which release will be a good one, and stick with it for a while. This can be hard to judge.
To be honest, at this stage I'm fairly tired with upgrading code to meet the current version of standards compliance under different compilers. It's kind of satisfying the first few times, but that doesn't last. I prefer to work on functionality.
A.
This is a perfectly valid/interesting post, given the slashdot community's focus on patent rights and renderering technology. Why the fuck is it moderated troll?
I mean, I'm getting the impression most posters are just here to go, "ooh, look at the pretty fishies."
A.
Perhaps it's more to fend off the competition. Also, bear in mind that they make Renderman, which is used by other companies.
Pixar is a pretty nasty company when it comes to patenting graphics techniques that are either obvious or that have been previously presented at academic talks or conferences. They're pretty quick to use such patents, too.
It's probably best to steer clear of any technique they're using. (Subdivision surfaces etc.)
A.
Microsoft's millions of SLASHTROTURFING minions are RUTHLESSLY suppressing the shocking truth REVEALED in the parent post!
The truth is out there! Or maybe in here. Whatever.
A.
> I find that the best program is almost always the shortest.
I disagree.
If you are working on a large code-base, in conjunction with many other people, this is untrue. Over-abstracting or over-generalizing code has the effect of making that code much harder to parse, and much harder to maintain.
Code abstraction is akin to compression. You make it more difficult for people to decrypt (decompress) what the code is doing, unless the abstraction you're using is a common one that everyone can be expected to know. You are also making the code much less flexible. If code is heavily abstracted, any given required modification is more likely to render the abstraction invalid and require a complete rewrite of the code.
This is the real reason Lisp/Scheme/SML/whatever the current flavour of the month is has never caught on in industry in a large way. PL people like these languages because they make it much easier to produce short, dense, powerful code, and this appeals to the mathematician within. There needs to be more human factors research in PL. Or perhaps they should just change the name of the research area. Much PL work is theoretically valuable, even if it is at best orthogonal to producing a good programming language.
Abstraction is a fine tool when applied carefully, but too many people in academia, especially those who have never written a program over one thousand lines, misjudge where the balance point is.
My friend, meet the game "industry". Game industry: meet my friend.
A.
P.S. for those also unfamiliar with the U.S. business environment, google reveals EIN to be "employer identification number."
> This isn't a friggen pissing contest...
Oh, so very very wrong =)
A.
Pffft. Double-click is not a Microsoftism. It may well have originated with the original Mac Finder, though I could be wrong on that.
It's puzzling how you could be so ignorant, unless you're too young to remember the 80s. Which may not be such a bad thing.
A.
> Then I get to spend the rest of my time posting on /. on how wrong you clueless Linux users are about ... and rebooting machines.
>
> the state of the Windows world.
A.
No, wait, it was Christopher Marlowe.
A.
Sorry, but you left out the bit about Shakespeare's plays really being written by Francis Bacon.
A.
This guy is a professor? I really hope this article isn't representative of his academic work. I know economics is the dismal science, but really:
> The one big piece of evidence that I didn't have
> when we talked before was a half-year year 2002
> number [that appears to indicate a 9.8 percent
> decrease in album sales.] There has to be a
> caveat in here, which is that I don't know
> if this number is correct. It's a half-year
> number that I saw in USA Today, from SoundScan.
So his entire paper is based on some number he read in his morning news. Umm. And then:
> If [we] assume that half the computer owners have
> CD burners -- a number that I've seen -- you just
> double the decline that's already occurred.
I don't know where to start with this. I've read better logic on crop circle web pages.
Where's the science?
Hmm. A quick web search reveals he spends most of his time defending Microsoft.
A.
Heard it a hundred times. It was relevant five years ago, but not today.
Repeat after me: the price/performance curve for commodity graphics hardware is outstripping that for general CPUs. Graphics hardware's "Moore's Law" is winning.
Now, never make the mistake of assuming that an exponential curve will increase indefinitely -- they're all S-curves in the end -- but while it does, movie and PC graphics quality are not both increasing at the same rate. So assuming that the gap between them isn't going to at the very least close a bit is wishful thinking.
A.
Score 1 "Offtopic"? WTF?
Or, are these people too thick to get sarcasm?
A.
Private sources. However, this is a public thread I found that summarizes things pretty well (apologies in advance for size):
F -8 &oe=UTF-8&safe=off&threadm=s08z3z7kxi.fsf%40envy.g raphics.cornell.edu&rnum=6&prev=/groups%3Fq%3Dexlu na%26hl%3Den%26lr%3D%26ie%3DUTF-8%26oe%3DUTF-8%26s afe%3Doff%26selm%3Ds08z3z7kxi.fsf%2540envy.graphic s.cornell.edu%26rnum%3D6
http://groups.google.com/groups?hl=en&lr=&ie=UT
With Pixar patenting everything they can lay their hands on, including obvious methods of how you'd use subdivision surfaces in production, and deep shadow maps, the future for rendering R&D looks a little grim unless you're a big company.
A.
Again, it's not true. BMRT has been killed as a result of the Pixar lawsuit. Nothing to do with NVIDIA, who have if anything acted as "rescue squad".
A.
Not quite.
Pixar has successfully prevented ExLuna from selling their core rendering product, a cutting-edge renderer based on the renderman standard, and also made them yank BMRT, by suing them for patent infringement, and threatening them with all sorts of other nasty legal stuff. (ExLuna is/was a company run by Larry Gritz, amongst other ex-Pixarites, and Larry was the author of BMRT.)
You won't see BMRT again.
NVIDIA has since bought ExLuna for their personnel and expertise.
Yay Pixar. It's amazing how far they've taken that bogus distributed sampling patent.
A.
You should have capitalised Nazi, too.
A.
Is there anyone awake at the editorial helm? Hello?
Bugger the headstones -- this story is so old it's six foot under itself.
A.
> First to use the Hurd.
Enough already -- you've convinced me.
A.
Of course, soon the attacking side will nick another trick from rugby; take the free kick as quickly as possible, so the defenders don't have a chance to get back 10. =)
A.
No thanks. I used to watch hockey (go Pens), but have given up on the game as of this season. With the constant clutching and grabbing and fouling, it must be the dullest sport in the world. (Okay, I exaggerate -- there's always golf.) The only winning strategy is interfering with or screening the goalie while someone shoots from outside. Yawn.
I should make it clear I'm talking about the North American version. The larger-rinked European game is probably more fun.
A.
> Pixar tried to make high-speed hardware for years, and we always found it to be a losing game. I
> wrote microcode for one of these beasts, a parallel bitslice engine that inspired today's MMX
> instructions. We could not keep up with the development of vanilla CPUs, and the CPUs ended
> up being more cost effective.
It is indeed hard keeping up with anything in the consumer space, due to economies of scale.
On the other hand, graphics hardware is now squarely in that space. Many of the points you make are very reasonable (e.g. general purpose vs. specialized for real-time), but I wonder if the consumer-oriented nature of current graphics hardware might not render them moot.
My PhD advisor, Paul Heckbert, told tales of writing microcode for the same beasts; it was obviously a scarring experience =). (Interestingly, Paul is currently working for NVIDIA.)
A.