I looked up "computer" once in an 1828 Noah Webster dictionary reprint. The definition was "one who computes." I think mathematicians employed assistants much as engineers employ technicians, to do the time-consuming grunt work.
Johannes Kepler had this problem (trying to emperically verify his theories, I suppose) and tried to build a fairly complex mechanical calculator, but failed to finish it before he died. This was long before Babbage.
How about punched tape? It would be more compact, I think. Probably best to use some kind of long-lived plastic rather than paper (mylar maybe?). The holes could be really close together and maybe 32 or 64 bits across the width of the tape.
Won't people basically be switching to DVD burners pretty soon?
OTOH I just got an awesome deal on a Yamaha CD-RW on ebay... my old burner died, and I wanted to put off spending the money. I like it; sure is a lot better than my old burner, being SCSI and all.
What about that? What about adaptive delta coding, is that the same thing as ADPCM? I just know that ADPCM is already supported in WAV files and has been for years. Maybe it has strings attached though.
Even gzip might be able to achieve 50% compression, right? I wonder why they can't achieve higher ratios?
It was a good deal too at half-price. I forget exactly what price though; maybe $6 or 7? And I only found one string; the rest had sold out already. They are shaped like stars, run on 120V and seem to work well but I haven't put them on anything yet. This was the day after Christmas at a Kmart in Scottsdale, AZ.
I noticed the first time I tried that the quality was a bit bad considering the bitrate... commercial VCDs look better than my first SVCD. But I didn't realize it's because the software needs improvement... I guess it's like MP3 encoding, having to get the psychoacoustic model right (or in this case the what, psychovisual model?) I just figured it was because I first used MainActor to generate motion-JPEG, and then converted that to MPEG, so there were two opportunities for artifacts to appear, and the MPEG engine had a chance to amplify the artifacts from the JPEG. Now, with Kino in good enough shape that I don't need MainActor, I can keep it in DV format until the MPEG encoding.
Anyway since last summer, new versions of everything came out, and for whatever reason my latest disc looks a lot better, IMO.
And does it come with scripts to automatically check everything against the debsums database and email discrepancies like integrit does?
They could use the debsums to "prime" the integrit db at installation time, so there is not that window of opportunity to trojan the binaries between the installation and the first integrit scan.
(I use integrit on debian but was not aware of this debsums thing. Sounds cool.)
Pretty soon they ought to be doing comparisons involving Linux too, not just Mac and Windows. Kino is beginning to seriously kick ass. It's now adequate for all my home video purposes (transferring camcorder video, editing and titling, making SVCDs).
That was my first Linux server back in '94. (Yes really 5 megs memory, I had 4 1M SIMMs and 4 256K SIMMs) One 40M Seagate 251 for/usr and some old 30 meg full-height beast for/. I had an RLL card and I think maybe I had converted the drives to RLL by then, but I forget exactly... I also had an ISP which was tolerant of my 24/7 connection, and even gave me a subdomain (ecloud.goodnet.com) so I could run a web server. Ah the memories...
Sure but that's a kludge and costs more. It would be so trivial to output RGB via an option in the DSP firmware, using the same output jacks. That's probably what my old Samsung does.
If the GUI resembles Steelcase cabinets with folders and files in them, and this is a restrictive view of information management, and it's also a very close view of how the OS manages information, then doesn't it follow that the very filesystem is itself too restrictive?
I don't think you can completely achieve his vision as just a pretty face on top of a conventional OS. (Oddly he seems to think he can...) You need something to replace the filesystem.
As a start, imagine ReiserFS could be accessed in multiple ways with multiple "views". One view is hierarchical, one is chronological. The chronological view can show you every file, recursively and without regard for folder structure, in order of last modification. Now that would be useful wouldn't it? It's practically a log of what things you did in what order, and requires no separate information store to achieve... the required metadata is already there in a conventional filesystem. You could achieve it with "find" plus some kind of sort-by-date, but that wouldn't be efficient; better to have a by-date index built into the filesystem, and new system calls to acccess it.
But that's just one example. Such a filesystem still doesn't have built-in version control, like say VMS, or like Xanadu. Nor does it have relational features like BeOS or the AS/400, nor a resource fork like the old MacOS filesystem. (But maybe those could be alternate views, if a filesystem can have "n" views of different kinds with different possible operations in each.) And the usual paradigm in most OSes assumes that the hierarchy of directories is the bottom-level, barebones, required way of organizing everything. If you want a relational database, you can do it but the whole thing will consist of files on top of the filesystem. An object database can be built but it will be stored in one or more files too. Organizing MP3s is so hard that most jukeboxes use relational databases to keep track of the metadata that isn't in the MP3s themselves, or that is too slow to search by reading the ID tags on each file one by one. This all seems pathetic to me; and it's so arbitrary that hierarchy should be king. Why are humans so comfortable with needing a tree, that they can't think of other ways as being equally valid rather than afterthought add-ons? And really, I think chronological organization should definitely be right up there with hierarchy, and you need a way to do intersections: find me the objects which were modified during this time period, in this folder and its descendants, which have relationships with such-and-such person or project, or something like that. If such methods of finding things were sufficiently efficient and powerful, maybe we wouldn't need the plain old tree at all anymore. It's just hard to come up with an alternative solution which fits all situations. Perhaps Xanadu comes closest to being universal but it's still mostly vaporware, and its chronological organization abilities also need improvement.
Pen computing has had plenty of incarnations. Windows for Pen really did exist; I have a tablet PC which has this OS, the Dauphin. It's based on Windows 3.1, but you can write anywhere on the screen and it will be recognized and inserted as text wherever the cursor happens to be sitting at the time. There is a program called Microsoft Notebook which comes with it; it has virtual pages, and on them you can both draw pictures and write handwriting. It can tell which is which. Pictures can be left as-is, or "cleaned up" to neat lines and boxes and ellipses automatically. Text is converted to text objects automatically (with normal fonts rather than your handwriting). Keep in mind this is all from the early 90's, and the Dauphin is a 486 with 6 megs of memory and a 40 meg hard drive.
IBM had a handwriting enabled DOS called PenDOS which ran on this machine too, I think, but I've never seen it.
Go was the company that made software for that HP PDA, wasn't it? I remember that. It was around the same time as the 100lx, 200lx etc. and those were more popular.
Then there's the Newton. Wasn't "digital ink" invented there? As far as I know the Newton had no real faults, it was just too expensive, and that bastard Jobs killed it before the economies of scale could make it cheap enough like Palms. I sure never managed to get one when they were hip, but I could probably afford it now that they are getting cheap on ebay, and see what all the fuss is about.
And most of today's PDAs use a form of pen computing. Just that using a Grafitti pad is much more restrictive than in-place handwriting recognition, or being able to choose to keep and show it in "digital ink" form rather than in styled text.
You can find a lot of old pen-based Windows systems on ebay, such as the Dauphins, the Concerto, the IBM 730T (I have one of those too), the GRiD machines (such as the 2260 and 2270 - I have some of those too) etc.
Anyway the point is Pen Computing is not new, nor is it vaporware; but I find that it's awkward - I'd much rather type than write. I hated writing so much in school that being a geek is such an escape for me... I don't use pens much anymore for the most part. Using a pen with a computer, for text input, is the last thing I want, especially as many mistakes as handwriting recognition usually entails. But using it for drawing is the intriguing part, I think. Software which combines that with a more efficient way to enter text will be very cool, but I don't think we have found the right combination of ideas yet. Most likely this new Tablet PC movement is just another round of the same old thing, improved and better integrated (like using digital ink everywhere, not just in a standalone Microsoft Notebook application), but still using a pen for input is not ideal for all situations and never will be. If it's to be portable for quick-and-dirty everyday small stuff, a Newton has a better form factor; and for real work, I cannot do without a keyboard until something equally efficient is invented, and handwriting ain't it.
It's important to have a keyboard, but they will, won't they? If you can use it as a tablet but also pop up the screen to reveal a keyboard underneath, that's better than a laptop, IMO. (I'm thinking of the GRiD 2260/2270 design...)
What I really really want to see is a hack to make the component video outputs VGA compatible. When this is possible I can upgrade my old Samsung which has this feature, but otherwise is very buggy and slow. I use it with an Electrohome projector and a wonderful electronic switcher that handles SVGA, NTSC, RGB (including VGA from a PC so I can play back movie files or watch XawTV) but not that wacko YCbCr stuff. Why can't they just use RGB for TVs anyway, it's much more natural, one channel for each electron gun in a conventional CRT...
Nothing is what it seems. Most experiences are disappointing. Coding the same thing more than once is a bore; and each time, when you're finished the company dies and takes their proprietary software down with them, and you have nothing except the money you probably already spent, and your hard work never sees the light of day, or is properly rewarded if it does.
Probably the best advice is if you have food on the table and a roof with a computer and a 'net connection under it, be happy; and if you like coding so much, get your thrills from releasing open source. A job is generally just a job - to support your habits.
Maybe if you had an advanced degree you could work in a cool research lab, but otherwise the only real opportunities will be those you create for yourself.
I successfully made a couple SVCDs; but I started with digital video (from a digital 8mm camcorder) and played them on a DVD player (a Sampo model - they are probably the most versatile and hackable players). I didn't have any sync problems, but the video quality left a bit to be desired (it looks short of VHS quality to me, even though encoded at SVCD bitrate, and thus I can only get 1/2 hour on a disc). I did subtitles and of course they were the worst part to encode. I used MainActor for editing and subtitles; but video output from Kino is just as good, it's only that Kino is a bit under-featured at this point, so I got MA as a stopgap (and don't recommend it - it crashes a lot). SVCDs and VCDs can have chapters like DVDs, so it's best to put each chapter in its own AVI file (besides, AVI files have length limits - 2 gigs or something like that).
Here are my notes about how I made one disc:
edit-??.avi are exported from maseq using AVI-mjpeg, default quality, 720x480, 30fps, interlace A, de-interlaced
I wrote a script for this encoding project and went to bed; it took a long time.
I wondered if I got some quality degradation by exporting from MA in motion-JPEG format, rather than keeping it in native DV format, and then encoding to MPEG. Ideally some of the JPEG frames would just directly become keyframes in the MPEG output; but in this case I was scaling too, so that's not possible. Anyway most of the output formats in MainActor for Linux have bugs, and MJPEG happened to work well.
By the way, speaking of mounting issues... the kitchen in our house came with a knife-shelf which pulls down from the underside of the cabinets. I never used it so I mounted the Audrey there. (I had to modify its case a bit and also cut a hole in the bottom of the shelf, for the lump in the back with all the connectors to stick through, so that it can lie flat on its back on the shelf). And there's enough space left on the shelf for the keyboard, too. So it's out of the way of grease splatters most of the time, unless you're actually using it and thus have the shelf pulled down.
Eventually I want to figure out how to use a CueCat with it (it has USB ports and I have a USB CueCat) and also a OneWire network adapter so I can control the sprinkler system in the garden window with it.
As for having messy fingers, you just have to be disciplined enough not to use the touchscreen without washing up first. Better have the recipe scrolled into position before you start getting dirty.
I agree, Audrey is awesome. I bought 4 of them. My web server has scripts to control X-10 lights, and a voicemail system (vgetty plus some playback and management scripts), and a few recipes; the Audreys are just decent little touchscreen web browsers. They can play audio, so I can check voicemail from various places around the house. There is also an MP3 player but the quality isn't good enough really. I also wrote a client-server caller ID system; the server sends out UDP broadcasts and the Audrey client pops up a little dialog when the phone rings. The Audrey runs QNX, and you can get the development platform for free (as in beer).
Yeah but you can't just go live in Canada on a whim, right? You'd have to immigrate, and from what I hear that's tough (part of why their population is so small).
There's a much more interesting use for this...
on
WiFi Triangulation
·
· Score: 2
...than security.
Remember this? They used an ultrasonic echo-location system to build a spatial "mouse" which could be used to turn posters on walls into "smart posters" (click here to turn the lights on and off, etc.) and also to track users within their lab, so that your phone calls are forwarded to the phone nearest you, etc. At the time, I thought, how redundant... they need ultrasound for tracking and an RF system of some sort to transmit "clicks". Why not just use a wireless network and come up with a triangulation method to find the location of the WiFi device using its own emissions. Well now it's been done. So it should be possible to use a PDA with a WiFi card as that magic 3D mouse thing. Imagine having location-relevant UIs for things: as you walk down a hall you get light-switch controls on your screen for nearby rooms, a map, the meeting schedule for the nearby conference room, reminders about stuff you need to do while you are in this area of the building, instant-messaging informs your colleagues that you are nearby, etc.
Of course for smart-poster purposes, the resolution ought to be better (1 meter isn't good enough) but perhaps that could be improved.
I think in the future location tracking will usually have 2 tiers: outside you use GPS, and inside buildings you use radio-triangulation of some kind. It will be a sort of standard eventually. Because you need higher resolution indoors, for various reasons. And since buildings don't move, the building triangulation system can tell you precisely where the building's "origin" is in lat/long space, so you would still be using GPS-style coordinates, just with greater accuracy in indoor situations. Instead of being deprived due to the fact that GPS signals don't penetrate well enough, you actually get better quality.
Anybody else tired of security always being in the limelight? Yes we need that kind of geek very much, but fundamentally their job is a lot more boring than what's going on in the research labs... And these security "mine's bigger than yours" wars are getting almost as annoying as the MS hate-fest, or the Apple hate-fest of a decade ago.
Slashdot Mirror
I looked up "computer" once in an 1828 Noah Webster dictionary reprint. The definition was "one who computes." I think mathematicians employed assistants much as engineers employ technicians, to do the time-consuming grunt work.
Johannes Kepler had this problem (trying to emperically verify his theories, I suppose) and tried to build a fairly complex mechanical calculator, but failed to finish it before he died. This was long before Babbage.
How about punched tape? It would be more compact, I think. Probably best to use some kind of long-lived plastic rather than paper (mylar maybe?). The holes could be really close together and maybe 32 or 64 bits across the width of the tape.
Won't people basically be switching to DVD burners pretty soon?
OTOH I just got an awesome deal on a Yamaha CD-RW on ebay... my old burner died, and I wanted to put off spending the money. I like it; sure is a lot better than my old burner, being SCSI and all.
What about that? What about adaptive delta coding, is that the same thing as ADPCM? I just know that ADPCM is already supported in WAV files and has been for years. Maybe it has strings attached though.
Even gzip might be able to achieve 50% compression, right? I wonder why they can't achieve higher ratios?
I can load that page but the links don't seem to work, and the price is rather unbelievable. What's going on here?
In your wallet it's more likely to collect lint than dust, actually.
It was a good deal too at half-price. I forget exactly what price though; maybe $6 or 7? And I only found one string; the rest had sold out already. They are shaped like stars, run on 120V and seem to work well but I haven't put them on anything yet. This was the day after Christmas at a Kmart in Scottsdale, AZ.
I noticed the first time I tried that the quality was a bit bad considering the bitrate... commercial VCDs look better than my first SVCD. But I didn't realize it's because the software needs improvement... I guess it's like MP3 encoding, having to get the psychoacoustic model right (or in this case the what, psychovisual model?) I just figured it was because I first used MainActor to generate motion-JPEG, and then converted that to MPEG, so there were two opportunities for artifacts to appear, and the MPEG engine had a chance to amplify the artifacts from the JPEG. Now, with Kino in good enough shape that I don't need MainActor, I can keep it in DV format until the MPEG encoding.
Anyway since last summer, new versions of everything came out, and for whatever reason my latest disc looks a lot better, IMO.
And does it come with scripts to automatically check everything against the debsums database and email discrepancies like integrit does?
They could use the debsums to "prime" the integrit db at installation time, so there is not that window of opportunity to trojan the binaries between the installation and the first integrit scan.
(I use integrit on debian but was not aware of this debsums thing. Sounds cool.)
Pretty soon they ought to be doing comparisons involving Linux too, not just Mac and Windows. Kino is beginning to seriously kick ass. It's now adequate for all my home video purposes (transferring camcorder video, editing and titling, making SVCDs).
The name is cymballic.
Did anybody capture it?
That was my first Linux server back in '94. (Yes really 5 megs memory, I had 4 1M SIMMs and 4 256K SIMMs) One 40M Seagate 251 for /usr and some old 30 meg full-height beast for /. I had an RLL card and I think maybe I had converted the drives to RLL by then, but I forget exactly... I also had an ISP which was tolerant of my 24/7 connection, and even gave me a subdomain (ecloud.goodnet.com) so I could run a web server. Ah the memories...
Sure but that's a kludge and costs more. It would be so trivial to output RGB via an option in the DSP firmware, using the same output jacks. That's probably what my old Samsung does.
I don't think you can completely achieve his vision as just a pretty face on top of a conventional OS. (Oddly he seems to think he can...) You need something to replace the filesystem.
As a start, imagine ReiserFS could be accessed in multiple ways with multiple "views". One view is hierarchical, one is chronological. The chronological view can show you every file, recursively and without regard for folder structure, in order of last modification. Now that would be useful wouldn't it? It's practically a log of what things you did in what order, and requires no separate information store to achieve... the required metadata is already there in a conventional filesystem. You could achieve it with "find" plus some kind of sort-by-date, but that wouldn't be efficient; better to have a by-date index built into the filesystem, and new system calls to acccess it.
But that's just one example. Such a filesystem still doesn't have built-in version control, like say VMS, or like Xanadu. Nor does it have relational features like BeOS or the AS/400, nor a resource fork like the old MacOS filesystem. (But maybe those could be alternate views, if a filesystem can have "n" views of different kinds with different possible operations in each.) And the usual paradigm in most OSes assumes that the hierarchy of directories is the bottom-level, barebones, required way of organizing everything. If you want a relational database, you can do it but the whole thing will consist of files on top of the filesystem. An object database can be built but it will be stored in one or more files too. Organizing MP3s is so hard that most jukeboxes use relational databases to keep track of the metadata that isn't in the MP3s themselves, or that is too slow to search by reading the ID tags on each file one by one. This all seems pathetic to me; and it's so arbitrary that hierarchy should be king. Why are humans so comfortable with needing a tree, that they can't think of other ways as being equally valid rather than afterthought add-ons? And really, I think chronological organization should definitely be right up there with hierarchy, and you need a way to do intersections: find me the objects which were modified during this time period, in this folder and its descendants, which have relationships with such-and-such person or project, or something like that. If such methods of finding things were sufficiently efficient and powerful, maybe we wouldn't need the plain old tree at all anymore. It's just hard to come up with an alternative solution which fits all situations. Perhaps Xanadu comes closest to being universal but it's still mostly vaporware, and its chronological organization abilities also need improvement.
IBM had a handwriting enabled DOS called PenDOS which ran on this machine too, I think, but I've never seen it.
Go was the company that made software for that HP PDA, wasn't it? I remember that. It was around the same time as the 100lx, 200lx etc. and those were more popular.
Then there's the Newton. Wasn't "digital ink" invented there? As far as I know the Newton had no real faults, it was just too expensive, and that bastard Jobs killed it before the economies of scale could make it cheap enough like Palms. I sure never managed to get one when they were hip, but I could probably afford it now that they are getting cheap on ebay, and see what all the fuss is about.
And most of today's PDAs use a form of pen computing. Just that using a Grafitti pad is much more restrictive than in-place handwriting recognition, or being able to choose to keep and show it in "digital ink" form rather than in styled text.
You can find a lot of old pen-based Windows systems on ebay, such as the Dauphins, the Concerto, the IBM 730T (I have one of those too), the GRiD machines (such as the 2260 and 2270 - I have some of those too) etc.
Anyway the point is Pen Computing is not new, nor is it vaporware; but I find that it's awkward - I'd much rather type than write. I hated writing so much in school that being a geek is such an escape for me... I don't use pens much anymore for the most part. Using a pen with a computer, for text input, is the last thing I want, especially as many mistakes as handwriting recognition usually entails. But using it for drawing is the intriguing part, I think. Software which combines that with a more efficient way to enter text will be very cool, but I don't think we have found the right combination of ideas yet. Most likely this new Tablet PC movement is just another round of the same old thing, improved and better integrated (like using digital ink everywhere, not just in a standalone Microsoft Notebook application), but still using a pen for input is not ideal for all situations and never will be. If it's to be portable for quick-and-dirty everyday small stuff, a Newton has a better form factor; and for real work, I cannot do without a keyboard until something equally efficient is invented, and handwriting ain't it.
It's important to have a keyboard, but they will, won't they? If you can use it as a tablet but also pop up the screen to reveal a keyboard underneath, that's better than a laptop, IMO. (I'm thinking of the GRiD 2260/2270 design...)
What I really really want to see is a hack to make the component video outputs VGA compatible. When this is possible I can upgrade my old Samsung which has this feature, but otherwise is very buggy and slow. I use it with an Electrohome projector and a wonderful electronic switcher that handles SVGA, NTSC, RGB (including VGA from a PC so I can play back movie files or watch XawTV) but not that wacko YCbCr stuff. Why can't they just use RGB for TVs anyway, it's much more natural, one channel for each electron gun in a conventional CRT...
Nothing is what it seems. Most experiences are disappointing. Coding the same thing more than once is a bore; and each time, when you're finished the company dies and takes their proprietary software down with them, and you have nothing except the money you probably already spent, and your hard work never sees the light of day, or is properly rewarded if it does.
Probably the best advice is if you have food on the table and a roof with a computer and a 'net connection under it, be happy; and if you like coding so much, get your thrills from releasing open source. A job is generally just a job - to support your habits.
Maybe if you had an advanced degree you could work in a cool research lab, but otherwise the only real opportunities will be those you create for yourself.
Here are my notes about how I made one disc:
I wondered if I got some quality degradation by exporting from MA in motion-JPEG format, rather than keeping it in native DV format, and then encoding to MPEG. Ideally some of the JPEG frames would just directly become keyframes in the MPEG output; but in this case I was scaling too, so that's not possible. Anyway most of the output formats in MainActor for Linux have bugs, and MJPEG happened to work well.
Eventually I want to figure out how to use a CueCat with it (it has USB ports and I have a USB CueCat) and also a OneWire network adapter so I can control the sprinkler system in the garden window with it.
As for having messy fingers, you just have to be disciplined enough not to use the touchscreen without washing up first. Better have the recipe scrolled into position before you start getting dirty.
Some links: Here's my caller ID thing and audreyhacking.com will tell you everything else you need to know.
Yeah but you can't just go live in Canada on a whim, right? You'd have to immigrate, and from what I hear that's tough (part of why their population is so small).
Remember this? They used an ultrasonic echo-location system to build a spatial "mouse" which could be used to turn posters on walls into "smart posters" (click here to turn the lights on and off, etc.) and also to track users within their lab, so that your phone calls are forwarded to the phone nearest you, etc. At the time, I thought, how redundant... they need ultrasound for tracking and an RF system of some sort to transmit "clicks". Why not just use a wireless network and come up with a triangulation method to find the location of the WiFi device using its own emissions. Well now it's been done. So it should be possible to use a PDA with a WiFi card as that magic 3D mouse thing. Imagine having location-relevant UIs for things: as you walk down a hall you get light-switch controls on your screen for nearby rooms, a map, the meeting schedule for the nearby conference room, reminders about stuff you need to do while you are in this area of the building, instant-messaging informs your colleagues that you are nearby, etc.
Of course for smart-poster purposes, the resolution ought to be better (1 meter isn't good enough) but perhaps that could be improved.
I think in the future location tracking will usually have 2 tiers: outside you use GPS, and inside buildings you use radio-triangulation of some kind. It will be a sort of standard eventually. Because you need higher resolution indoors, for various reasons. And since buildings don't move, the building triangulation system can tell you precisely where the building's "origin" is in lat/long space, so you would still be using GPS-style coordinates, just with greater accuracy in indoor situations. Instead of being deprived due to the fact that GPS signals don't penetrate well enough, you actually get better quality.
Anybody else tired of security always being in the limelight? Yes we need that kind of geek very much, but fundamentally their job is a lot more boring than what's going on in the research labs... And these security "mine's bigger than yours" wars are getting almost as annoying as the MS hate-fest, or the Apple hate-fest of a decade ago.