My startup, Draftastic, provides VCS-like tools for ordinary text. We have a lot of stuff Etherpad doesn't, like Markdown formatting and a permissions system, and our paragraph granularity (instead of line or character) avoids a lot of annoying problems with writing over each other. Also, we work with JavaScript off. And our free demo isn't slashdotted;).
Draftastic is a web-based collaborative editor that avoids lock contention issues and works without JavaScript, among other good things.
It's free for a single document. Paid accounts get more documents, a permission system, and so on.
(Not OSS, but built using mostly open-source technology. We've contributed a few patches already, and are hoping to find other ways to "give back to the community".)
> DSLRs do typically allow long exposures and such
Sure, but you're not expected to do a 30-second exposure at f/2 in sunlight. You might as well just use a white bucket fill in Photoshop. Given a sensor and ISO, proper exposure is more or less defined as a certain amount of light hitting the sensor, so we can come up with a number of photons that's close to every reasonable exposure. In other words, a typical one-second shot doesn't have 1000 times the light hitting the sensor as a 1/1000 second shot.
With light and settings such that 1/1000 s is the proper exposure, a 30-second movie without the shutter closing is obviously the equivalent light of 30,000 exposures. That's as many as I've put on my 350D in the 16 months I've had it, and probably a good fraction of the expected body lifetime. If you take one such movie a day for a year, that's easily five times the light a high-end DSLR sensor is ever expected to see.
But again, I don't actually have evidence for the fading-dye theory. I just heard it once, did some estimations, and thought it seemed plausible.
I've heard that the dyes used on the higher-quality DSLR sensors tend to fade slightly in strong light, so video would degrade the color quality over time. A five-minute video in strong sunlight might let as much light fall on the sensor as thousands of still frames (assuming there was no shutter action -i.e., the light is continuous and chopped into video frames internally). I haven't been able to verify this, though.
I'm reading the article article you linked to, and it seems to support the greenhouse gas theory. For instance, in section 5 (split over pages 2 and 3):
In terms of the causes of the increase in ocean heat content we believe that the long-term trend as seen in these records is due to the increase of greenhouse gases in the Earth's atmosphere
I can't find any mentions of volcanism. Please point them out or provide a better source.
Remember when images came along and the Internet hung because it was designed for text? And then, only a few years later, music crashed all the transoceanic links? That sucked.
If the anti-video-watching PSAs fail, the only solution will be paying somewhat higher monthly fees for somewhat better service. And that just wouldn't make sense. Telcos would have to... I don't know... lay new cables? Is that even possible?
Heck, 6 megapixels is fine for most print use if you don't crop. But I think photographers will be asking for more resolution well up into the tens of megapixels because it gives you the freedom to throw away half the frame and keep something sharp. If you have tremendous resolution, a wide-angle lens gives zoom-like detail, and an f/1.8 50mm lens is $70 whereas an f/4 500mm is $5,800. For most photographers, the ones whose systems cost less than houses, it's cheaper to get a denser sensor than a longer lens and it will probably stay that way for a while. Also, bit depth is almost certainly going way up.
So yeah, "Pictures will go upto 10 megapixels but it will stop there" is nuts. Heaps of people will find ways to use, and slowly come to rely on, a doubling of megapixels just as easily as they do a doubling of core clock speeds.
Generic Linux may be more stable, but it's not as stable (or as predictable) as a realtime OS. Even one hang a year would be terrible for air traffic control, nuclear power plants, rocketry, etc. They would much rather have a slow, non-scalable system than one that had even a slim chance of screwing up.
Re:CodeTek versus Desktop Manager
on
Hacking Quartz
·
· Score: 1
The big volume control style white icon is quite annoying, but it can easily be turned off. In my experience, Desktop Manager is quite fast overall (if you turn off the eyecandy, of course), and works with X11. It still feels a little rough in places, but it's worth trying.
Hydra was renamed SubEthaEdit, and version 2.0 just came out. It's advertised as a collaborative text editor for pair programming etc., but it's also great for ordinary coding and config-editing. They rounded out the features in 2.0, making it a much better alternative to BBEdit, and fixed the syntax highlighting problems. It's closed-source, but free for personal use ($35/seat commercial base price). I recommend it -- and no, I'm not affiliated with it.
That's the option key. It does roughly the same kinds of things as "alt". The shortcut will be to hold down command ("Apple button") and option and hit the letter. Good luck.
From skimming the patent, it looks like they're patenting something vaguely like this:
<versions>
<version language='perl' interpreter='/usr/bin/perl'>
print("I am a banana!\n");
</version>
<version language='python' interpreter='/usr/bin/python'>
print 'I am a banana!'
</version>
</versions>
... in other words, using XML to keep several languages' versions of one script.
I don't really see the point. There are plenty of extremely portable languages, and what happens if the versions in the XML file fall out of synch? If someone edits the perl version but not the python version, you could be in trouble. Writing a non-trivial algorithm that works exactly the same in two completely different languages (if they weren't completely different, you wouldn't need to drag them both around) seems like more work than just using a portable language in the first place. I suppose it could be useful for keeping scripts across incompatible language versions -- you could have one script for $language v1 though v2.5, and one for all later versions.
Still, if I were using XML to make my code portable, I'd use Flare or something very much like it. Maybe I'm missing the point, but I think this patent is pretty weird.
Yes and no. Yes, there are some pretty silly claims about the value of good orthography. However, language is an extremly widely-supported interface, and you should be able to use it well.
Furthermore, although there are plenty of dyslexic geniuses, I think you'll find that there are even more who are good at writing. Off the top of my head, Einstein, Feynman, Knuth, Rob Pike, Alan Turing, J. B. S. Haldane, and Freeman Dyson are all very good writers. Being smart at the things they were famous for didn't keep them from being smart at explaining those thing.
Sure, you can just hire an editor. But you can just hire a software engineer. People do specialize, but that's no reason to completely ignore things outside your field, especially if they help it in the long term. Einstein would not have gone very far if he had been unable to articulate what Relativity meant to other mathematicians and physicists.
There's nothing wrong with learning to write outside "writing classes", provided you actually are learning. But entirely dismissing the study of writing is rather foolish. It's like dismissing touchtyping or code optimization or database design -- sure, you could learn it as you go along, but you'd better hope you're really good at it before you have to do a whole lot of it at once. You're always going to have to write something; you can't just duck it. If you use it well, it can be extremely useful in convincing people that your ideas are any good.
If nothing else, use a spellchecker. It makes you about twice as credible in one simple step.
Well, "/wolf/wired" is short enough that it would be ambiguous, but I agree completely with the sentiment. mod_rewrite all the way, man.
When every Amazon URI I can remember had "exec/obidos" in it, there's obviously some redundancy. ISBNs are near-universal book addresses, and (at the moment) they're 10 digits, but Amazon is making me look at "exec/obidos/tg/detail//03755 02904/qid=1068751824/sr=12-8/103-2810600-6302246" = 77 characters, and that's leaving out the "www." and the reasonable query string.
There are 36 alphanumeric characters (so we're not even counting/, =, etc.), and 36^77 is 68437251422602946514698890166699222907622494972995 29009512147947683368878710518963676135121221268151 26166485053049143296. Does Amazon have that many products? Does it have a millionth that many products? A billionth? A trillionth?
No? So why is it making me haul around that many characters? Web programming my foot.
We would have to keep it up, and each send a couple fake replies a week from different addresses. It's not like they would suddenly give up and never try again. And I'll bet the vast majority of spammers already have good anti-spam filters on their addresses -- after all, they're distributing them widely in cleartext.
Grassroots projects are fine and dandy, but one of the reasons we have a government is to enforce grassroots projects. Let's put down The Great Big Bathroom Book of Ayn Rand and wonder if we may not be able to solve this readily as a loose collection of autonomous agents.
Ignoring spam could work. If the signal/noise ratio is infinite, but a given spammer only gets one reply, they're still probably losing money. That's unlikely, but in principle I don't think it's any less likely than everyone spontaneously agreeing to consistently swamp smammers from a zillion throw-away addresses. (And have you thought about the bandwidth costs if we all did that?)
The kernel (Mach, in Mac OS X's case) doesn't matter much to a media player. Obviously things like hardware support and i/o latency make a difference, but more or less any modern OS will support something that supports MP3 playing. It probably wasn't a matter of a quick recompile, but iTunes works fine on Windows, which has a completely different hardware and low-level software architecture.
What makes the difference here is having a media architecture. iTunes floats on QuickTime, which Apple trusts to work really well with various audio codecs (and their DRM schemes). Besides that, it's (")just(") some GUI, network, security, and disc-burning code. QuickTime is the central issue here; Apple would not make iTunes for mpg123/ALSA/whatever.
But if you're asking "Why haven't they ported QuickTime to Linux yet?", I agree. I suppose they might be thinking like this:
We want to sell Macintoshes, and, if possible, software.
Let's make a killer app so people will want to buy Macs, and call it QuickTime.
But wait... if we make it Mac-only, not even Mac users will want to use it, because it'll be outnumbered by whatever Microsoft comes up with the stamp it out.
So let's port it, and make it a model citizen in the Windows envirnoment, so most everyone will be able to use it while associating it with Apple.
Port to Linux? Why? We want these people to moan and whine about not being able to watch.mov trailers, and talk themselves into buying a tibook. Obviously they're already vulnerable to Unix. Mwahahahrahra!
In other words, they have to port to Windows if they want it to survive at all. But they're the powerful ones in comparison to Linux, and they can just try to borg its users.
Disclaimers: (1) this is pure speculation, (2) I use Mac OS X considerably more than Linux these days, and (3) I'm feeding a troll.
Solar panels. They're already the best solution for most remote stuff in relatively sunny climates (navigational buoys, spacecraft), and they're still not very efficient (15%?).
Energy transmission by microwave or laser (e.g. for orbital solar power).
Floating seawater-cooled reactors. Don't laugh.
Passive or semi-passive stuff: tidal, geothermal, hydroelectric, weird-ass solar chimneys, etc.
Why muck around? Go for cold fusion. Yes, the most famous attempt was a fraud. Yes, it's not going to be ready tomorrow, even given a huge breakthrough. But the potential is amazing.
There are three basic kinds of power: grid power, which comes in bulk; portable fueled power, like a car engine; and embedded power, like a battery. All of could be a lot safer, cheaper, and cleaner. Happy research.
I don't agree that they're already more than twice as good.
However, I'd say your estimates are if anything conservative. 100dpi might be low, but there's no reason each 'page' has to be paper-page-sized. I've found that half a letter/A4 is just right for writing on and carrying.
For note-taking, there's no way you need 24-bit color. 8-bit should be fine, and maybe even 4-bit grayscale. The PNG compression should be more than satisfactory. Textbooks, being stored as mostly text and vector graphics instead of mostly bitmap, would presumably be a lot more compact. Figure a little overhead for OS and software to render, search, recognize handwriting, etc., but well under 0.5GiB.
The only problem I have with your estimates is that you don't count batteries. I assume tablet power consumption is quite low, but by the time I filled up all 33k pages, I'd probably have gone through a few cells (and spent a bit on charging them). Other than that, I agree with your figures.
But that's not the issue for me, because I don't need 33k pages stored. I don't even need 1k pages stored. Anything over perhaps 300 pages should be in something more indexable than JPEG/PNG, so we're talking about intelligent vectorization again, which tablets can't do and aren't likely to do for quite a while as far as I can tell.
So raw storage is just not the issue for me. It could store a billion pages, and it still wouldn't be significatly more useful to me without a keyboard, other good I/O, and plenty of speed and memory... which would make it a laptop, which I already own. And even my laptop can't reliably turn my scribbles into equations, flowcharts, indented pseudocode, "commented-out" doodles, etc., so I don't care.
I'm happy with nothing filling the gap between a pencil and Mac OS. Someday an amazing tablet my convince me otherwise. Not today.
Paper and pen(cil) are very good for taking sketch-like notes. If you'd just convert raster sketch-notes taken on a tablet to MathML (e.g.) anyway, you might as well do the same thing working from paper. In other words, there's no extra convenience in recording notes digitally if you're going to do a computationally difficult/impossible transform on them before it matters whether they're digital.
I think any tablet would have to have an extremely good equation-recognition system before most people would find it useful for taking notes. Even then it would be competing with pen and paper for speed and flexibility of interface. I wouldn't consider paying for something that just gave me what I'd get if I scanned my analog notes.
Tablets might be a good idea someday, but I don't think even Apple can do it with (what little I know of) current natural-graphics-recognition technology. Wake me up when they're twice as good as pen and paper and cost less than three times as much.
"Technology" does not mean "something new and exciting"; it means "a way of doing something". Think of "technique".
It's blatantly obvious that paper was once a technology, and that it still is. Perhaps it would have been worthwhile to remind us that paper was once a new or high technology.
Slashdot Mirror
(Self-link.)
My startup, Draftastic, provides VCS-like tools for ordinary text. We have a lot of stuff Etherpad doesn't, like Markdown formatting and a permissions system, and our paragraph granularity (instead of line or character) avoids a lot of annoying problems with writing over each other. Also, we work with JavaScript off. And our free demo isn't slashdotted ;).
(Warning: self-link.)
Draftastic is a web-based collaborative editor that avoids lock contention issues and works without JavaScript, among other good things.
It's free for a single document. Paid accounts get more documents, a permission system, and so on.
(Not OSS, but built using mostly open-source technology. We've contributed a few patches already, and are hoping to find other ways to "give back to the community".)
> DSLRs do typically allow long exposures and such
Sure, but you're not expected to do a 30-second exposure at f/2 in sunlight. You might as well just use a white bucket fill in Photoshop. Given a sensor and ISO, proper exposure is more or less defined as a certain amount of light hitting the sensor, so we can come up with a number of photons that's close to every reasonable exposure. In other words, a typical one-second shot doesn't have 1000 times the light hitting the sensor as a 1/1000 second shot.
With light and settings such that 1/1000 s is the proper exposure, a 30-second movie without the shutter closing is obviously the equivalent light of 30,000 exposures. That's as many as I've put on my 350D in the 16 months I've had it, and probably a good fraction of the expected body lifetime. If you take one such movie a day for a year, that's easily five times the light a high-end DSLR sensor is ever expected to see.
But again, I don't actually have evidence for the fading-dye theory. I just heard it once, did some estimations, and thought it seemed plausible.
I've heard that the dyes used on the higher-quality DSLR sensors tend to fade slightly in strong light, so video would degrade the color quality over time. A five-minute video in strong sunlight might let as much light fall on the sensor as thousands of still frames (assuming there was no shutter action -i.e., the light is continuous and chopped into video frames internally). I haven't been able to verify this, though.
I'm reading the article article you linked to, and it seems to support the greenhouse gas theory. For instance, in section 5 (split over pages 2 and 3):
I can't find any mentions of volcanism. Please point them out or provide a better source.
That's for prose. For academic prose, try Joseph Williams' Style.
Remember when images came along and the Internet hung because it was designed for text? And then, only a few years later, music crashed all the transoceanic links? That sucked.
... I don't know ... lay new cables? Is that even possible?
If the anti-video-watching PSAs fail, the only solution will be paying somewhat higher monthly fees for somewhat better service. And that just wouldn't make sense. Telcos would have to
Heck, 6 megapixels is fine for most print use if you don't crop. But I think photographers will be asking for more resolution well up into the tens of megapixels because it gives you the freedom to throw away half the frame and keep something sharp. If you have tremendous resolution, a wide-angle lens gives zoom-like detail, and an f/1.8 50mm lens is $70 whereas an f/4 500mm is $5,800. For most photographers, the ones whose systems cost less than houses, it's cheaper to get a denser sensor than a longer lens and it will probably stay that way for a while. Also, bit depth is almost certainly going way up.
So yeah, "Pictures will go upto 10 megapixels but it will stop there" is nuts. Heaps of people will find ways to use, and slowly come to rely on, a doubling of megapixels just as easily as they do a doubling of core clock speeds.
Generic Linux may be more stable, but it's not as stable (or as predictable) as a realtime OS. Even one hang a year would be terrible for air traffic control, nuclear power plants, rocketry, etc. They would much rather have a slow, non-scalable system than one that had even a slim chance of screwing up.
The big volume control style white icon is quite annoying, but it can easily be turned off. In my experience, Desktop Manager is quite fast overall (if you turn off the eyecandy, of course), and works with X11. It still feels a little rough in places, but it's worth trying.
Hydra was renamed SubEthaEdit, and version 2.0 just came out. It's advertised as a collaborative text editor for pair programming etc., but it's also great for ordinary coding and config-editing. They rounded out the features in 2.0, making it a much better alternative to BBEdit, and fixed the syntax highlighting problems. It's closed-source, but free for personal use ($35/seat commercial base price). I recommend it -- and no, I'm not affiliated with it.
That's the option key. It does roughly the same kinds of things as "alt". The shortcut will be to hold down command ("Apple button") and option and hit the letter. Good luck.
The rovers run vxWorks, a commercial OS, as reported on Slashdot two weeks ago.
From skimming the patent, it looks like they're patenting something vaguely like this:
<versions><version language='perl' interpreter='/usr/bin/perl'>
print("I am a banana!\n");
</version>
<version language='python' interpreter='/usr/bin/python'>
print 'I am a banana!'
</version>
</versions>
... in other words, using XML to keep several languages' versions of one script.
I don't really see the point. There are plenty of extremely portable languages, and what happens if the versions in the XML file fall out of synch? If someone edits the perl version but not the python version, you could be in trouble. Writing a non-trivial algorithm that works exactly the same in two completely different languages (if they weren't completely different, you wouldn't need to drag them both around) seems like more work than just using a portable language in the first place. I suppose it could be useful for keeping scripts across incompatible language versions -- you could have one script for $language v1 though v2.5, and one for all later versions.
Still, if I were using XML to make my code portable, I'd use Flare or something very much like it. Maybe I'm missing the point, but I think this patent is pretty weird.
Yes and no. Yes, there are some pretty silly claims about the value of good orthography. However, language is an extremly widely-supported interface, and you should be able to use it well.
Furthermore, although there are plenty of dyslexic geniuses, I think you'll find that there are even more who are good at writing. Off the top of my head, Einstein, Feynman, Knuth, Rob Pike, Alan Turing, J. B. S. Haldane, and Freeman Dyson are all very good writers. Being smart at the things they were famous for didn't keep them from being smart at explaining those thing.
Sure, you can just hire an editor. But you can just hire a software engineer. People do specialize, but that's no reason to completely ignore things outside your field, especially if they help it in the long term. Einstein would not have gone very far if he had been unable to articulate what Relativity meant to other mathematicians and physicists.
There's nothing wrong with learning to write outside "writing classes", provided you actually are learning. But entirely dismissing the study of writing is rather foolish. It's like dismissing touchtyping or code optimization or database design -- sure, you could learn it as you go along, but you'd better hope you're really good at it before you have to do a whole lot of it at once. You're always going to have to write something; you can't just duck it. If you use it well, it can be extremely useful in convincing people that your ideas are any good.
If nothing else, use a spellchecker. It makes you about twice as credible in one simple step.
Well, "/wolf/wired" is short enough that it would be ambiguous, but I agree completely with the sentiment. mod_rewrite all the way, man.
When every Amazon URI I can remember had "exec/obidos" in it, there's obviously some redundancy. ISBNs are near-universal book addresses, and (at the moment) they're 10 digits, but Amazon is making me look at "exec/obidos/tg/detail//03755 02904/qid=1068751824/sr=12-8/103-2810600-6302246" = 77 characters, and that's leaving out the "www." and the reasonable query string.
There are 36 alphanumeric characters (so we're not even counting /, =, etc.), and 36^77 is 68437251422602946514698890166699222907622494972995 29009512147947683368878710518963676135121221268151 26166485053049143296. Does Amazon have that many products? Does it have a millionth that many products? A billionth? A trillionth?
No? So why is it making me haul around that many characters? Web programming my foot.
We would have to keep it up, and each send a couple fake replies a week from different addresses. It's not like they would suddenly give up and never try again. And I'll bet the vast majority of spammers already have good anti-spam filters on their addresses -- after all, they're distributing them widely in cleartext.
Grassroots projects are fine and dandy, but one of the reasons we have a government is to enforce grassroots projects. Let's put down The Great Big Bathroom Book of Ayn Rand and wonder if we may not be able to solve this readily as a loose collection of autonomous agents.
Ignoring spam could work. If the signal/noise ratio is infinite, but a given spammer only gets one reply, they're still probably losing money. That's unlikely, but in principle I don't think it's any less likely than everyone spontaneously agreeing to consistently swamp smammers from a zillion throw-away addresses. (And have you thought about the bandwidth costs if we all did that?)
The kernel (Mach, in Mac OS X's case) doesn't matter much to a media player. Obviously things like hardware support and i/o latency make a difference, but more or less any modern OS will support something that supports MP3 playing. It probably wasn't a matter of a quick recompile, but iTunes works fine on Windows, which has a completely different hardware and low-level software architecture.
What makes the difference here is having a media architecture. iTunes floats on QuickTime, which Apple trusts to work really well with various audio codecs (and their DRM schemes). Besides that, it's (")just(") some GUI, network, security, and disc-burning code. QuickTime is the central issue here; Apple would not make iTunes for mpg123/ALSA/whatever.
But if you're asking "Why haven't they ported QuickTime to Linux yet?", I agree. I suppose they might be thinking like this:
In other words, they have to port to Windows if they want it to survive at all. But they're the powerful ones in comparison to Linux, and they can just try to borg its users.
Disclaimers: (1) this is pure speculation, (2) I use Mac OS X considerably more than Linux these days, and (3) I'm feeding a troll.
I don't know about really obscure stuff, but we could take another look at something obvious: power.
There are three basic kinds of power: grid power, which comes in bulk; portable fueled power, like a car engine; and embedded power, like a battery. All of could be a lot safer, cheaper, and cleaner. Happy research.
The Blue Marble images go up to 21,600 by 21,600 -- about 450 megapixels.
I don't agree that they're already more than twice as good.
However, I'd say your estimates are if anything conservative. 100dpi might be low, but there's no reason each 'page' has to be paper-page-sized. I've found that half a letter/A4 is just right for writing on and carrying.
For note-taking, there's no way you need 24-bit color. 8-bit should be fine, and maybe even 4-bit grayscale. The PNG compression should be more than satisfactory. Textbooks, being stored as mostly text and vector graphics instead of mostly bitmap, would presumably be a lot more compact. Figure a little overhead for OS and software to render, search, recognize handwriting, etc., but well under 0.5GiB.
The only problem I have with your estimates is that you don't count batteries. I assume tablet power consumption is quite low, but by the time I filled up all 33k pages, I'd probably have gone through a few cells (and spent a bit on charging them). Other than that, I agree with your figures.
But that's not the issue for me, because I don't need 33k pages stored. I don't even need 1k pages stored. Anything over perhaps 300 pages should be in something more indexable than JPEG/PNG, so we're talking about intelligent vectorization again, which tablets can't do and aren't likely to do for quite a while as far as I can tell.
So raw storage is just not the issue for me. It could store a billion pages, and it still wouldn't be significatly more useful to me without a keyboard, other good I/O, and plenty of speed and memory ... which would make it a laptop, which I already own. And even my laptop can't reliably turn my scribbles into equations, flowcharts, indented pseudocode, "commented-out" doodles, etc., so I don't care.
I'm happy with nothing filling the gap between a pencil and Mac OS. Someday an amazing tablet my convince me otherwise. Not today.
Paper and pen(cil) are very good for taking sketch-like notes. If you'd just convert raster sketch-notes taken on a tablet to MathML (e.g.) anyway, you might as well do the same thing working from paper. In other words, there's no extra convenience in recording notes digitally if you're going to do a computationally difficult/impossible transform on them before it matters whether they're digital.
I think any tablet would have to have an extremely good equation-recognition system before most people would find it useful for taking notes. Even then it would be competing with pen and paper for speed and flexibility of interface. I wouldn't consider paying for something that just gave me what I'd get if I scanned my analog notes.
Tablets might be a good idea someday, but I don't think even Apple can do it with (what little I know of) current natural-graphics-recognition technology. Wake me up when they're twice as good as pen and paper and cost less than three times as much.
"Technology" does not mean "something new and exciting"; it means "a way of doing something". Think of "technique".
It's blatantly obvious that paper was once a technology, and that it still is. Perhaps it would have been worthwhile to remind us that paper was once a new or high technology.