I own about a half-dozen macs. I know what zapping the PRAM officially does. I also know that people in comp.sys.mac.system will recommend it for anything from "machine won't boot" to "SCSI termination problems". I don't know whether it has side-effects that we don't really know about.
I'm not exactly a Mac expert, but I do write about 'em a lot.
For what it's worth, one of the examples that I had in mind when I wrote this was the conversion from xchat to xchat2.
xchat could redraw its entire display in much less than a screen refresh. xchat2 takes visible time to redraw the screen, and if the system is at all loaded, it can take several seconds.
The smoothness of fonts is, at least in this case, a total non-issue to me; I'm using a 1600x1200 screen and individual pixels are nearly invisible to me.
I even used Dell as an example (either here or in the book I'm working on) of a company I won't buy from because I don't like being asked to pick a customer type before I can buy products.
I think this is why Canon's winning the high-end printing market.
Ink for a high-end Canon inkjet (say, the i9100 or i9900) is a plastic cartridge with a sponge in it and a bunch of ink, and runs you $12.
Ink for HP printers has a CPU, a dozen forms of copy protection, region coding, and a team of ninjas to defend against the possibility that you'll refill it.
Ironically, the reason it costs enough to be worth trying to bypass them is that their costs are huge, because the ink cartridge is full of special stuff. Canon's actually selling ink, and there's no one using these printers for whom it would be worth the time to try to "cheat" and not buy their reasonably-priced ink.
I've been using the paid version of GC for ages, and I love it to bits. If you haven't gotten it, GET IT! It's the best toy ever. In particular, the 3D objects are just plain FUN. Note that you can specify hsv or rgb values for color, so you can have parametric color shading of elaborate 3d objects. There's an entire family of interesting objects you can get with, say, summing sin(x)+sin(y)+sin(z), and animating this.
And no, I'm not a shill; I'm just a fanatic who was very happy to hear that there's finally an OS X native version of graphing calculator. I love this program.
The problem isn't the proliferation of standards; it's the proliferation of non-standards.
Standards help us. They make us have to learn one thing, not ten or twenty. I mean, there's still lots of things to learn... But we can at least learn a given technology once.
It'd be nice if they'd put up a note on the project page to that effect. Currently, any attempt to get links for ATI TV in and out support goes to their long-defunct page, and there's not much reference to how to get it working in current X.
Of course, to actually WATCH the TV input, you need software. Contrary to popular belief, Cyberlink PowerVCR is teh sux0r, and no amount of fidgeting was ever able to get it to synchronize the signal correctly; their support staff said to "check that my video driver was current", and I eventually gave up and got a refund. Capturix Video Suite worked fine, though.
The GATOS and related projects which were once working on this seem to have silently disintegrated without touching XF86 4.4.x, although it could be that there's some kind of support and I just have no clue where to find documentation. But... No external dongle, and it's a laptop with video in.
Not to say it's COMMON, mind you, but it does exist.
(The A31p was the Best Laptop Ever, and I wish IBM would sell something at least COMPARABLE to it, but nothing in their current lineup can match the three-spindle monster machine. Curious tidbit: Although it's not in the official specs, an A31p can have 2GB of memory!)
Why did RealNetworks send out millions of pieces of spam to totally arbitrary addresses, with a return address which claimed to be Maria Cantwell, but was actually thrown away unread?...
Hey, guys, I mean, music stores are neat, but Real Networks is one of the largest spammers in the world, having proudly bragged about having a 53,000,000 address list back when that was probably better than 10% of the people who regularly checked their email. They once proudly proclaimed that they were not going to take bounces off their list, because a lot of those bounces were from spam filters, and they didn't want to take people off their list because "hey, they could be bouncing us as spam, but the address is still valid".
I guess... I'm not seeing much difference here from things we can already do, like having a window be a single large block and telling the GPU where to render it.
I don't think there's as much room for change here as you seem to be thinking of, although I guess that depends somewhat on how efficient the 2D acceleration in X is these days.
That's more efficient than having the CPU do all the rendering, but... It's not exactly running the whole thing on the GPU. The UI has to manipulate a fair number of polygons to do that; I'm not sure this is much of an improvement over having a 2D image and telling the GPU to render that.
I think the idea of a GPU having a specific opcode for "a window" is a little unlikely. That level of specialization would be spending die space on a feature whose performance is plenty fast already; I doubt it'll happen.
I don't really buy the claim that the GPU will be "drawing the UI". I mean, obviously, the GPU is what actually renders the pixels, but someone's still telling it what to do.
I'm aware that the people working on this might not be able to do the other kinds of work we need done... But it's still not much for "news". This stuff doesn't need to be part of X; window managers have been doing it for years.
I suspect that even a hypothetical perfect use would still use at least some CPU cycles.
Anyway, I'm well aware that some people like eye-candy. I just wonder whether it's the best thing to be putting a lot of development effort into when the base functionality isn't worked out.
I'm aware that it's not all the same, but... Seems to me that "X can run slower" isn't much of a feature to be bragging about.
Anyway, to me, it seems like the default should be to pass through key sequences which aren't being used by the UI. Nothing in the UI seems to care much about alt-keys, that I can detect.
But... Yes, the feature of passing through the keysym for the key I hit should be "normal". That would be a canonical example of a normal feature.
This is why most systems have a standard for what key you use to send out-of-band commands, and if that dead key isn't down, everything you send is some kind of input. Works great.
"If you're unable to type international characters in the same way in both xterm and gaim/other gtk apps, your system is broken."
Yup.
That's very nice.
But... Shift-Alt-5 gets me a nice yen symbol in xterm, and Nothing Happens in gaim.
Telling me it's "broken" doesn't help much; these are stock installs. So... Yeah, it's broken. That's sorta my point; it would be nice to see stuff like this being fixed, rather than gratuitous shinies. (And yes, I do sometimes send in code to open source projects. I know the deal.)
Slashdot Mirror
It is indeed an April Fools joke.
It also shows some neat things about vectorizing non-trivial examples.
I own about a half-dozen macs. I know what zapping the PRAM officially does. I also know that people in comp.sys.mac.system will recommend it for anything from "machine won't boot" to "SCSI termination problems". I don't know whether it has side-effects that we don't really know about.
I'm not exactly a Mac expert, but I do write about 'em a lot.
For what it's worth, one of the examples that I had in mind when I wrote this was the conversion from xchat to xchat2.
xchat could redraw its entire display in much less than a screen refresh. xchat2 takes visible time to redraw the screen, and if the system is at all loaded, it can take several seconds.
The smoothness of fonts is, at least in this case, a total non-issue to me; I'm using a 1600x1200 screen and individual pixels are nearly invisible to me.
While we're doing Cranky User columns, I wrote about this long ago:
Pigeonholed!
I even used Dell as an example (either here or in the book I'm working on) of a company I won't buy from because I don't like being asked to pick a customer type before I can buy products.
VINDICATED!
I think this is why Canon's winning the high-end printing market.
Ink for a high-end Canon inkjet (say, the i9100 or i9900) is a plastic cartridge with a sponge in it and a bunch of ink, and runs you $12.
Ink for HP printers has a CPU, a dozen forms of copy protection, region coding, and a team of ninjas to defend against the possibility that you'll refill it.
Ironically, the reason it costs enough to be worth trying to bypass them is that their costs are huge, because the ink cartridge is full of special stuff. Canon's actually selling ink, and there's no one using these printers for whom it would be worth the time to try to "cheat" and not buy their reasonably-priced ink.
Well, you can do it lots of ways. I even said the threading was "an experiment". :)
Personally, if I were doing this again, I probably wouldn't use threading for it.
When I first saw it, it said "Estimated arrival by January 22", now it says "3-4 weeks". I assume there was a rush early on.
It'll be interesting seeing whether it can be easily set up for TV out.
I've been using the paid version of GC for ages, and I love it to bits. If you haven't gotten it, GET IT! It's the best toy ever. In particular, the 3D objects are just plain FUN. Note that you can specify hsv or rgb values for color, so you can have parametric color shading of elaborate 3d objects. There's an entire family of interesting objects you can get with, say, summing sin(x)+sin(y)+sin(z), and animating this.
And no, I'm not a shill; I'm just a fanatic who was very happy to hear that there's finally an OS X native version of graphing calculator. I love this program.
Who or what uses "LM hash"? It'd be nice to have an example application or two.
As noted by the Economist's endorsement, and also as commented in my blog entries about the election, there's serious issues here, which are ignored by many Republicans.
The problem isn't the proliferation of standards; it's the proliferation of non-standards.
Standards help us. They make us have to learn one thing, not ten or twenty. I mean, there's still lots of things to learn... But we can at least learn a given technology once.
Wow!
:)
That's incredibly stupid.
Sounds like someone skimmed a few marketing brochures and wrote about the results at length, without doing any actual research.
FWIW, I use all three on and off, NetBSD being my favorite, and OpenBSD being my second-favorite. FreeBSD is a little too dodgy for my tastes.
It'd be nice if they'd put up a note on the project page to that effect. Currently, any attempt to get links for ATI TV in and out support goes to their long-defunct page, and there's not much reference to how to get it working in current X.
YMMV. My experience with ATI drivers, on both my laptop and my desktop, has been that they're consistently pretty decent.
My ThinkPad A31p has video inputs.
Of course, to actually WATCH the TV input, you need software. Contrary to popular belief, Cyberlink PowerVCR is teh sux0r, and no amount of fidgeting was ever able to get it to synchronize the signal correctly; their support staff said to "check that my video driver was current", and I eventually gave up and got a refund. Capturix Video Suite worked fine, though.
The GATOS and related projects which were once working on this seem to have silently disintegrated without touching XF86 4.4.x, although it could be that there's some kind of support and I just have no clue where to find documentation. But... No external dongle, and it's a laptop with video in.
Not to say it's COMMON, mind you, but it does exist.
(The A31p was the Best Laptop Ever, and I wish IBM would sell something at least COMPARABLE to it, but nothing in their current lineup can match the three-spindle monster machine. Curious tidbit: Although it's not in the official specs, an A31p can have 2GB of memory!)
>Its common sense that the more easily available
>the firearm is, the more likely it is that it gets
>into the wrong hands.
And the more likely it gets into the right hands.
The less legally available a firearm is, the more likely it is that *only* the bad guys have it.
Why did RealNetworks send out millions of pieces of spam to totally arbitrary addresses, with a return address which claimed to be Maria Cantwell, but was actually thrown away unread? ...
Hey, guys, I mean, music stores are neat, but Real Networks is one of the largest spammers in the world, having proudly bragged about having a 53,000,000 address list back when that was probably better than 10% of the people who regularly checked their email. They once proudly proclaimed that they were not going to take bounces off their list, because a lot of those bounces were from spam filters, and they didn't want to take people off their list because "hey, they could be bouncing us as spam, but the address is still valid".
IANMTU. These people are scary.
I guess... I'm not seeing much difference here from things we can already do, like having a window be a single large block and telling the GPU where to render it.
I don't think there's as much room for change here as you seem to be thinking of, although I guess that depends somewhat on how efficient the 2D acceleration in X is these days.
That's more efficient than having the CPU do all the rendering, but... It's not exactly running the whole thing on the GPU. The UI has to manipulate a fair number of polygons to do that; I'm not sure this is much of an improvement over having a 2D image and telling the GPU to render that.
GPUs aren't magic.
I'm not sure I want them happening in all WMs...
I think the idea of a GPU having a specific opcode for "a window" is a little unlikely. That level of specialization would be spending die space on a feature whose performance is plenty fast already; I doubt it'll happen.
I don't really buy the claim that the GPU will be "drawing the UI". I mean, obviously, the GPU is what actually renders the pixels, but someone's still telling it what to do.
I'm aware that the people working on this might not be able to do the other kinds of work we need done... But it's still not much for "news". This stuff doesn't need to be part of X; window managers have been doing it for years.
I suspect that even a hypothetical perfect use would still use at least some CPU cycles.
Anyway, I'm well aware that some people like eye-candy. I just wonder whether it's the best thing to be putting a lot of development effort into when the base functionality isn't worked out.
I'm aware that it's not all the same, but... Seems to me that "X can run slower" isn't much of a feature to be bragging about.
Anyway, to me, it seems like the default should be to pass through key sequences which aren't being used by the UI. Nothing in the UI seems to care much about alt-keys, that I can detect.
But... Yes, the feature of passing through the keysym for the key I hit should be "normal". That would be a canonical example of a normal feature.
This is why most systems have a standard for what key you use to send out-of-band commands, and if that dead key isn't down, everything you send is some kind of input. Works great.
"If you're unable to type international characters in the same way in both xterm and gaim/other gtk apps, your system is broken."
Yup.
That's very nice.
But... Shift-Alt-5 gets me a nice yen symbol in xterm, and Nothing Happens in gaim.
Telling me it's "broken" doesn't help much; these are stock installs. So... Yeah, it's broken. That's sorta my point; it would be nice to see stuff like this being fixed, rather than gratuitous shinies. (And yes, I do sometimes send in code to open source projects. I know the deal.)
Er.
You're talking to the guy who wrote an article called "Drowning in Aqua".
I think my dislike of user interface CPU-wasters is pretty generic.