The best thing about Xbox is that it won't change. Ever.
Yes, boys and girls, this is the shining city on the hill brought to you by your good friends at Microsoft®: pay no attention to the rats, or to the raw sewage flowing in the streets, or the bands of armed thugs lurking in dark alleys to rob or murder innocent passersby. Without Microsoft® none of these great innovations(TM) would have been possible.
Why don't you download and burn a CD (or a couple of CDs) of the sources for the distribution CD's you are currently selling? You would then have the source in your own posession for the required three years.
IANAL, but are you really responsible for providing the source code if all you are doing is burning exact copies of someone elses distro? Can't you just point anyone that comes to you for the source back to the original distributor?
Just one further note, I got these links by using a new-fangled invention called a search engine on another moder dohicky called the world-wide-web. You aught to give it a try sometime, rather than having other folks do all the work for you.
The Palm IIIe, which costs about $150, has a 16MHz Dragonball processor (MC68K derivative) along with 2MB of RAM and a couple of serial ports (one of which is an IrDA port). There is a fair amount of hacking information available on the web for the Palm devices, and you can get a stripped down version of Linux for it.
The only big drawback I can see, for the IIIe specifically, is that it doesn't have any flash memory, so you can't burn your own software into it, as you can with the IIIx or IIIxe. Still, you could always write your software for PalmOS, since there is a fair amount of information and software available for that.
If you are just going to doing robotics projects, however, you may want to look at some of the single chip microcontrollers, like the 683xx, 68HC12 or 68HC16 families from Motorola. Some of those have both RAM and EEPROM built-in, as well as a handfull of usefull peripherals (serial ports, ADC, DAC, etc.) and cost only a few (to a few tens of) dollars each.
From a few back-of-the-envelope calculations I can see that a minimal 1TB setup would cost at least $7200, for the drives alone (this is using 18 60GB IDE drives without any kind of redundancy) A more reliable solution would be to use SCSI drives in a RAID configuration, which would need 1.5 times as many drives in order to deliver some redundancy. Now we are looking at 27 drives and 4 SCSI busses, which comes to about $40,000, again just for the drives.
MacOS supports both multithreading and multiprocessing using separate interfaces.
The multithreading support is a hold-over from the pre-PowerPC days. It provides cooperatively scheduled threads and may only work for M68K code. The multiprocessing support is from the period immediately preceeding the return of Steve Jobs. It was introduced to support the PowerMac 9500/ 180MP and 9600/ 200MP, which had 2 processors, but it can be used on single processor machines as well. Threads created using multiprocessing services are preemtively scheduled, unlike almost everything else (aside from vertical retrace routines) on MacOS.
The downside of all three systems (thread manager, multiprocessing services, and vertical retrace manager) is that you are restricted in which OS calls are available to a thread/process/retrace routine. In general, you don't have access to certain parts of your applications memory space, you are not allowed to do anything which might cause memory to be allocated or rearranged[1] and you don't have much, if any, direct access to the GUI (you can calculate all kinds of stuff, but you can't interract with the user).
What this all comes down to is that you can make use of multiple processors using current versions of MacOS (all versions after at least 7.6 support the multiprocessing services) but the MacOS itself doesn't make use of more than one processor: you need to write your application to specifically use multiple processors. However, you can run multiprocessing targeted applications on single processor machines without any problem.
when MacOS finds that there isn't enough space to allocate a requested block of memory, it will try to rearrange stuff in memory to make enough room. Remember, MacOS dates from a time before transparant memory management in hardware
Back in the early ninties (1992 or so) before my first programming job, I set as my professional goal: to become a C 'language lawyer'. (that is, the person in a programming team who understands the arcanum of a given programming language) Seven years later I am working at NASA with part of my job function being to provide C programming expertise to the other members of my team (I'm the only trained computer scientist on team, everyone else are physicists or methematicians). I have found that my goal was suprisingly easy to achieve and now need to formulate some kind of follow-up goal.
As a professional in any field, you should probably be reevaluating your goals on a periodic basis, as well as keeping an eye on how far you have progressed toward your current goal. You are likely to be working in your chosen field for far longer than it will take to achieve any specific goal, so you need to be thinking beyond the immediately upcoming goal.
You may have a small problem trying to project and image onto the windowshield, that will be visible to the driver, from behind the drivers seat. Since the windowshield is at a 45 degree angle, any image projected horizontally towards the windowshield from inside the vehicle will be reflected down towards the top of the dash board, not back towards the driver. (obviously this could be solved by mounting a vertical sheet of glass between the driver and the windowshield, but this poses some fairly severe safty hazzards and would probably make the vehicle no longer street legal)
A traditional 'heads-up' display is projected upwards (from a point approximately between the viewer's feet or knees) onto an inclined glass sheet which reflects the image back toward the viewer. In order to achieve a similar effect in an automobile you will need to mount the projector in the dashboard behind the instrument pannel, project upwards through the windowshield. With a small enough projector this shouldn't be a problem, but will require a fair amount of physical modification to the vehicle itself./P
Operating System Design: The XINU Approach by Douglas Comer (from Prentice Hall) was a really good book for me. It concentrates more on building a working operating system than on theoretical issues, which makes it a lot easier to read than some of the more academic texts. (Tenanbaum, Silberschatz and Galvin, etc.) There used to be several versions of the book, one each for the PDP-11, IBM-PC, and Macintosh, but you may not be able to find anything other than the IBM-PC version these days.
I also have an old copy of The Design and Implementation of the 4.3BSD UNIX Operating System by Leffler, McKusick, Karels and Quarterman, from Addison-Wesley, which was a pretty thorough coverage of unix concepts when it was written. There is likely to be a more up-to-date contemporary edition. (and you can probably find a small army of similar books, including the excelent Linux Kernel Internals by Beck, Böhme, Dziadzka, et. al. also from Addison-Wesley) Finally, you may want to pick up a copy of the anotated unix source code Lions' Commentary on UNIX 6th Edition, with Source Code by John Lions from Peer-to-Peer Communications.
If Intel had listened to consumers they would never have put the serial number there in the first place. The fact that Intel went ahead and marketed the P-III with the serial number feature intact means that they were just hoping to be able to bully consumers into accepting the 'feature', and they are only giving up on it becaue it has become clear that the market won't swallow it.
The boycott may have worked, but the fact that a boycott was even necessary underscores the fact that Intel is not a consumer friendly company.
I'm shocked that nobody else has mentioned Bruce Tognazzini's web site AskTog. He is the founder of the Apple Humam Interface Group and author of a number of good books on the subjects of software design and user interface design. His web site is wealth of usefull information on user interface design issues and asundry other topics.
roll your own TIVO replacement
on
Tivo Hacking?
·
· Score: 1
I have been meaning to turn my current Linux box into a TIVO-esqe device for some time now, but I haven't gotten around to getting all the requisite hardware to play nicely together (the curse of PC architectures: 'plug-and-play' doesn't)
I've already got a BT based video capture card and lots of spare disk space, but I'm wondering about capturing video to disk (I know I can play video directly into a window).
With modern processors you should be able to do some kind of video compression without specialized hardware, but it's not clear to me that the CPU has access to the video data (the BT card needs to be connected to the VGA feature port on the display card, which suggests to me that the BT card is writing directly to the display memory, rather than going through the CPU/main memory).
Another problem is the requirement for some pretty good compression. Raw, low quality video needs something like 4MB/sec, (320*200 pixels/frame * 2 bytes/pixel = 128KB/frame * 30 frames/sec = 3.8MB/sec) which will add up pretty rapidly (we're looking at 220MB/min or 13.6GB/hr) so compression is pretty much a necessity.
A reasonable video compression scheme will give you somewhere between 10:1 and 100:1 compression, which would mean that you could get between 10 and 100 hours of video into a 13GB disk, which is currently on the low end of the storage size spectrum.
Once you have the video data on disk you can easily burn it to CD (with a CD-RW) or back it up to tape (with a tape drive), which addresses the complaint about recording to VHS tape.
The BT card I have already has both input and output for standard video, so you wouldn't need a special card just to output to your TV. The rest of the system is a pretty standard bargain basement PC, which can be had for around $550 (US) and should have plenty of processing power and memory. The total system cost, retail, should come to less than $1000 (US) (I'd bet more in the $800 range), which is more than TIVO, but may be worth it in order to secure your privacy and continued consumer liberties.
The only open questions seem to be digital capture of the video data and access to a suitable codec. All the other (imporant) TIVO features can be taken care of either with cron jobs or perl scripts (and some suitable access to online TV listings, which are available in most metropolitan areas through the local nuewpaper's web-site).
I put LinuxPPC on my aging PowerMac clone but ended up reinstalling MacOS because I needed a few of the features I couldn't get (easily) under Linux:
multiple monitor support. I have two video cards and two monitors on the Mac and I hated to have one go to waste.
multimedia support. I've never been able to get the sound stuff working on my Linux box and XAnim doesn't hold a candle to the combination of RealMedia and QuickTime. At least not yet.
utility support. I needed something to sync my Palm IIIx to, and to download the pictures off my digital camera, and to access my old SCSI scanner, etc. I know I could do most of this stuff on Linux, but the effort required made it easier just to reinstall MacOS.
CD ROM burner support. I've got a CD-RW on the Linux box but have yet to successfully burn a CD on it. The Mac works like a champ, even if I can't do anything else on the box while a CD is burning.
productivity apps. I really need to be able to run a nice, simple productivity package now and then, and the Mac has better stuff for that than Linux, so far. The real problem here isn't that Linux doesn't have good stuff, but that I just prefer AppleWorks (nee ClarisWorks) to most anything else out there. If I could run AppleWorks on Linux (which may be possible under WINE, actually, since there is a WinNT version of AppleWorks) I'd be fine.
I am my wife's technical support, and she uses a Powerbook. When I need to fiddle with something on the Powerbook it is nice to have another MacOS system around to compare against.
finally, I find it usefull to have a second system, running completely different software than my primary system for various tasks (web design, running closed source shareware, porting my own software, etc.) and MacOS is much nicer on the nerves than Windows.
That said, I still spend about 80% of my time on the Linux box rather than the Mac. The Linux box is what I put all my effort and money into and the Mac is purely second string these days. I still enjoy the Mac, but Linux suits many of my needs (programming and general web surfing/email/news) better.
All-in-all, I wouldn't say I've so much "returned to the Dark Side" as "kept the Dark Side around as an amusing and occasionally usefull pet."
My advice to Netpliance is to consider several alternative products based on the iopener hardware but tailored to the Linux/hacker community (and priced at profit making levels):
an enhanced iopener with an ethernet port, unscrambled IDE header, and built-in splitter for the keyboard/mouse. Price this at an modest profit and market two versions for a) home users with DSL or cable modems, and b) Linux hackers who want a nice small network workstation.
a screenless iopener with two ethernet ports and a modem, running a Linux or *BSD firewall configuration, targeted at SOHO users. The market for home security appliances is just starting to open up and Netpliance should get in at the starting bell.
another screenless iopener with one ethernet port, a much faster processor, more memory, and a couple of large harddrives, marketed as an application/file server for the enhanced iopener on a SOHO network.
Each device shuld be essentially the same hardware as the current iopener with only minor variations in the manufacturing details. The devices each cover distinctly different segments within a single market (SOHO users and Linux/*BSD hackers) and could easly be sold at competitive and profitable prices.
The fact that Netpliance was able to offer the iopener at all indicates that it is only a matter of time before someone is offering these products to consumers. There is no good reason that Netpliance should be that someone.
For serious FP work the only reasonable choice is an Alpha: x86 just doesn't deliver very good FP performance. If you check out the Spec-95 scores (at http://www.spec.org/osg/cpu95/results/cfp95.html) for various systems you will find that, on SpecFP-95, Alphas out-perform almost all other systems by about 2:1 (with the exception of PA-RISC where the difference is only about 3:2). While x86 systems have been able to scale up the integer performance of the architecture, the FP performance has lagged considerably, probably do to the stack oriented FP register design (although it is possible that neither Intel nor AMD have seen FP performance as a prime motivator for sales, and hence haven't felt the need to push it in their designs).
Aho, Sethi & Ullman is a very good book, though some people think it's a bit unreadable. My personal favorite is Allen I. Holub's Compiler Design in C from Prentice Hall, which is more readable than Aho etc. without sacrificing any of the details. Another text, and the one that got me interested in compilers, is James E. Hendrix's A Small C Compiler from M&T Books which is also very readale by much less technical than Holub or Aho etc.
Under any circumstance, your prof is blowing a little bit of smoke up your ass with the talk of Aho, Sethi & Ullman being outdated just because it was written in 1988. Most of what we know about compiler design was known by the mid-sixties, if not before (some FORTRAN compilers were Really Good (R), even if many early C and Pascal compilers sucked). Although there was some progress made in the seventies and eighties, the state of the art hasn't changed all that much in the past ten years.
how about a TLD for jane citizen who isn't part of the network infrastructure, a governmental agency, commercial entity, educational institution, non-profit organization, or a branch of the military? I'd like to see someithing like.pri or.pub for private individuals.
As other posters have pointed out, this is an automatic process that will notice the attempt to store a duplicate file and use a symlink instead. There are some obvious problems with the scheme that I would REALLY like to see how they were solved:
if I modify a SIS stored file does that modify the copy other people see?
are there really all that many in the way of duplicate files on most servers? If so, then there wasn't all that much 'information' being stored on the server.
This reminds me of Ted Nelson's Xanadu project, which stored files as lists of references to other files. It would seem to me that storing references to common sub-ranges of files would be a real win, while this just sounds like a cheap hack that may look good on a feature list.
selling at a loss is called dumping and is illegal, at least in the U.S. The claims about Sony not making a profit off of the Playstation don't say that Sony was selling at a loss, just very near cost, and that any profit was made off the sale of games.
If MS were to try to sell a piece of hardware at a significant loss (and if the specs are at all reliable the loss would be quite significant) then they would be in even deeper doo-doo than they are already with the web-browser business.
By my (admittedly under informed) calculations, a box matching Nintendo Next's specs would cost almost $150 for parts alone, without the processor. Clever bookkeeping may be able to assign assembly, packaging and marketing costs away, but it would be very hard to cover up the raw parts costs.
What CPU is going to be out in the next six months (they'll need some lead time on the engineering) that will run at 1GHz and will cost less than $100? Something from Intel: unlikely due to cost. Something from IBM or Motorola: unlikely due to speed. Something from AMD: possible, but if they could do it I would think we would have heard more hype than this. I'm not inclined to believe a gaming rag, even on a good day, but this is just silly.
I ended up purchasing the LinuxPPC CD because I couldn't get the downloaded stuff to work right. (this was LinuxPPC 1999 with all the installer problems) but I'm not sorry to have spent a little cash for the CD. It arrived very rapidly and has been working flawlessly both at home (on an old PowerComputing PowerWave) and at work (on a PowerMac 7200/120). I plan to buy the 2000 version as well. (I was cheap and didn't get the free upgrade)
I don't want to look like an idiot or anything, but what does something like CodeWarrior give you that Emacs or XEmacs doesn't?
The main feature of CodeWarrior isn't the pretty IDE, it's the portable C++ GUI framework, PowerPlant. With PowerPlant you can write a single set of source code and recompile for any supported platform: MacOS, BeOS, Windows95/98/NT/etc, or X Windows. [X]Emacs doesn't do anything to address this, nor do most other IDEs, nor do any of the companies providing assorted compilers for different platforms.
Another feature of CodeWarrior, that was still in beta last time I owned a license on MacOS, was a Java compiler that produced binary executables instead of byte-code class files. While this may seem to defeat the primary purpose of Java (compile once, run anywhere) it produces much faster Java programs on the supported platforms.
A couple years ago I thought that having a version of CodeWarrior for Linux would be ideal. At the time I was doing almost all my programming on MacOS, with an eye toward porting to Windows. It seemed important, then, to be able to have a single code base (using PowerPlant) that would produce GUI executables for all target platforms.
Having an X version of PowerPlant would have been a big help, as would have been the ability to code in Java and compile straight to binary executables (something that could be done under CodeWarrior for MacOS)
These days, however, I'm doing pretty much all my coding for *nix/X Windows. With MacOS migrating to a Mach/BSD base I don't feel any need to write old-style MacOS versions of my code, and I figure that anyone using Windows can fend for themselves.
I think that CodeWarrior just missed their window of opportunity. A year or so ago there was a real need for good cross platform development tools. Now that it looks like the whole world will be running *nix in a few years, however, I don't think there is nearly as much interest (or need) for such things.
After thinking about it for a bit, I can see a Spiderman movie with a Raimi atmosphere being a Really Good Thing (TM). Raimi's TV productions have had just the right mix of light drama and outright zaniness for a Spidey movie. Besides, he can't do worse than Marvel group has been doing with their modern attempts at either animated or live action productions (anyone remember the Fantasitc Four movie that never even made it beyond the trailer? How about the Generation X abortion? And have you tried to watch any of the animated children's series? They are completely unwatchable: Even when you know what the plotlines are supposed to be, the cartoons are impossible to follow!)
DC has at least done reasonably well with their animated series, though the Batman movies went from acceptable to complete dreck. I'm very disappointed that Marvel has been squandering their best characters and plotlines in a frenzy of cheap marketing, obtuse media productions and rediculous revisionism.
Slashdot Mirror
From the second to last paragraph of the article
Yes, boys and girls, this is the shining city on the hill brought to you by your good friends at Microsoft®: pay no attention to the rats, or to the raw sewage flowing in the streets, or the bands of armed thugs lurking in dark alleys to rob or murder innocent passersby. Without Microsoft® none of these great innovations(TM) would have been possible.
Why don't you download and burn a CD (or a couple of CDs) of the sources for the distribution CD's you are currently selling? You would then have the source in your own posession for the required three years.
IANAL, but are you really responsible for providing the source code if all you are doing is burning exact copies of someone elses distro? Can't you just point anyone that comes to you for the source back to the original distributor?
- First, here are lecture notes from a college course on operating system design.
- Second, some more meterial from another university (it's not clear to me that this is from a course).
- Third, a terse document detailing broad set of features common to operating systems of different periods (also part of an operating sytems course).
- Fourth, another page, which seems to be part of college course, with a section on the history of operating systems.
- Fifth, a web-slideshow on the topic.
- And Finally, a smattering of other links to the same topic by even more authors: another lecture from a college course, chapter 3, section 1 from the book Introductory Information Protection by Fred Cohen & Associates, Operating Systems - Yesterday, Today, Tomorrow, and Evolution of Operating Systems User Interface Design
Just one further note, I got these links by using a new-fangled invention called a search engine on another moder dohicky called the world-wide-web. You aught to give it a try sometime, rather than having other folks do all the work for you.The Palm IIIe, which costs about $150, has a 16MHz Dragonball processor (MC68K derivative) along with 2MB of RAM and a couple of serial ports (one of which is an IrDA port). There is a fair amount of hacking information available on the web for the Palm devices, and you can get a stripped down version of Linux for it.
The only big drawback I can see, for the IIIe specifically, is that it doesn't have any flash memory, so you can't burn your own software into it, as you can with the IIIx or IIIxe. Still, you could always write your software for PalmOS, since there is a fair amount of information and software available for that.
If you are just going to doing robotics projects, however, you may want to look at some of the single chip microcontrollers, like the 683xx, 68HC12 or 68HC16 families from Motorola. Some of those have both RAM and EEPROM built-in, as well as a handfull of usefull peripherals (serial ports, ADC, DAC, etc.) and cost only a few (to a few tens of) dollars each.
From a few back-of-the-envelope calculations I can see that a minimal 1TB setup would cost at least $7200, for the drives alone (this is using 18 60GB IDE drives without any kind of redundancy) A more reliable solution would be to use SCSI drives in a RAID configuration, which would need 1.5 times as many drives in order to deliver some redundancy. Now we are looking at 27 drives and 4 SCSI busses, which comes to about $40,000, again just for the drives.
MacOS supports both multithreading and multiprocessing using separate interfaces.
The multithreading support is a hold-over from the pre-PowerPC days. It provides cooperatively scheduled threads and may only work for M68K code. The multiprocessing support is from the period immediately preceeding the return of Steve Jobs. It was introduced to support the PowerMac 9500/ 180MP and 9600/ 200MP, which had 2 processors, but it can be used on single processor machines as well. Threads created using multiprocessing services are preemtively scheduled, unlike almost everything else (aside from vertical retrace routines) on MacOS.
The downside of all three systems (thread manager, multiprocessing services, and vertical retrace manager) is that you are restricted in which OS calls are available to a thread/process/retrace routine. In general, you don't have access to certain parts of your applications memory space, you are not allowed to do anything which might cause memory to be allocated or rearranged[1] and you don't have much, if any, direct access to the GUI (you can calculate all kinds of stuff, but you can't interract with the user).
What this all comes down to is that you can make use of multiple processors using current versions of MacOS (all versions after at least 7.6 support the multiprocessing services) but the MacOS itself doesn't make use of more than one processor: you need to write your application to specifically use multiple processors. However, you can run multiprocessing targeted applications on single processor machines without any problem.
Back in the early ninties (1992 or so) before my first programming job, I set as my professional goal: to become a C 'language lawyer'. (that is, the person in a programming team who understands the arcanum of a given programming language) Seven years later I am working at NASA with part of my job function being to provide C programming expertise to the other members of my team (I'm the only trained computer scientist on team, everyone else are physicists or methematicians). I have found that my goal was suprisingly easy to achieve and now need to formulate some kind of follow-up goal.
As a professional in any field, you should probably be reevaluating your goals on a periodic basis, as well as keeping an eye on how far you have progressed toward your current goal. You are likely to be working in your chosen field for far longer than it will take to achieve any specific goal, so you need to be thinking beyond the immediately upcoming goal.
You may have a small problem trying to project and image onto the windowshield, that will be visible to the driver, from behind the drivers seat. Since the windowshield is at a 45 degree angle, any image projected horizontally towards the windowshield from inside the vehicle will be reflected down towards the top of the dash board, not back towards the driver. (obviously this could be solved by mounting a vertical sheet of glass between the driver and the windowshield, but this poses some fairly severe safty hazzards and would probably make the vehicle no longer street legal)
A traditional 'heads-up' display is projected upwards (from a point approximately between the viewer's feet or knees) onto an inclined glass sheet which reflects the image back toward the viewer. In order to achieve a similar effect in an automobile you will need to mount the projector in the dashboard behind the instrument pannel, project upwards through the windowshield. With a small enough projector this shouldn't be a problem, but will require a fair amount of physical modification to the vehicle itself./P
Operating System Design: The XINU Approach by Douglas Comer (from Prentice Hall) was a really good book for me. It concentrates more on building a working operating system than on theoretical issues, which makes it a lot easier to read than some of the more academic texts. (Tenanbaum, Silberschatz and Galvin, etc.) There used to be several versions of the book, one each for the PDP-11, IBM-PC, and Macintosh, but you may not be able to find anything other than the IBM-PC version these days.
I also have an old copy of The Design and Implementation of the 4.3BSD UNIX Operating System by Leffler, McKusick, Karels and Quarterman, from Addison-Wesley, which was a pretty thorough coverage of unix concepts when it was written. There is likely to be a more up-to-date contemporary edition. (and you can probably find a small army of similar books, including the excelent Linux Kernel Internals by Beck, Böhme, Dziadzka, et. al. also from Addison-Wesley) Finally, you may want to pick up a copy of the anotated unix source code Lions' Commentary on UNIX 6th Edition, with Source Code by John Lions from Peer-to-Peer Communications.
If Intel had listened to consumers they would never have put the serial number there in the first place. The fact that Intel went ahead and marketed the P-III with the serial number feature intact means that they were just hoping to be able to bully consumers into accepting the 'feature', and they are only giving up on it becaue it has become clear that the market won't swallow it.
The boycott may have worked, but the fact that a boycott was even necessary underscores the fact that Intel is not a consumer friendly company.
I'm shocked that nobody else has mentioned Bruce Tognazzini's web site AskTog. He is the founder of the Apple Humam Interface Group and author of a number of good books on the subjects of software design and user interface design. His web site is wealth of usefull information on user interface design issues and asundry other topics.
I have been meaning to turn my current Linux box into a TIVO-esqe device for some time now, but I haven't gotten around to getting all the requisite hardware to play nicely together (the curse of PC architectures: 'plug-and-play' doesn't)
I've already got a BT based video capture card and lots of spare disk space, but I'm wondering about capturing video to disk (I know I can play video directly into a window).
With modern processors you should be able to do some kind of video compression without specialized hardware, but it's not clear to me that the CPU has access to the video data (the BT card needs to be connected to the VGA feature port on the display card, which suggests to me that the BT card is writing directly to the display memory, rather than going through the CPU/main memory).
Another problem is the requirement for some pretty good compression. Raw, low quality video needs something like 4MB/sec, (320*200 pixels/frame * 2 bytes/pixel = 128KB/frame * 30 frames/sec = 3.8MB/sec) which will add up pretty rapidly (we're looking at 220MB/min or 13.6GB/hr) so compression is pretty much a necessity.
A reasonable video compression scheme will give you somewhere between 10:1 and 100:1 compression, which would mean that you could get between 10 and 100 hours of video into a 13GB disk, which is currently on the low end of the storage size spectrum.
Once you have the video data on disk you can easily burn it to CD (with a CD-RW) or back it up to tape (with a tape drive), which addresses the complaint about recording to VHS tape.
The BT card I have already has both input and output for standard video, so you wouldn't need a special card just to output to your TV. The rest of the system is a pretty standard bargain basement PC, which can be had for around $550 (US) and should have plenty of processing power and memory. The total system cost, retail, should come to less than $1000 (US) (I'd bet more in the $800 range), which is more than TIVO, but may be worth it in order to secure your privacy and continued consumer liberties.
The only open questions seem to be digital capture of the video data and access to a suitable codec. All the other (imporant) TIVO features can be taken care of either with cron jobs or perl scripts (and some suitable access to online TV listings, which are available in most metropolitan areas through the local nuewpaper's web-site).
I put LinuxPPC on my aging PowerMac clone but ended up reinstalling MacOS because I needed a few of the features I couldn't get (easily) under Linux:
That said, I still spend about 80% of my time on the Linux box rather than the Mac. The Linux box is what I put all my effort and money into and the Mac is purely second string these days. I still enjoy the Mac, but Linux suits many of my needs (programming and general web surfing/email/news) better.
All-in-all, I wouldn't say I've so much "returned to the Dark Side" as "kept the Dark Side around as an amusing and occasionally usefull pet."
My advice to Netpliance is to consider several alternative products based on the iopener hardware but tailored to the Linux/hacker community (and priced at profit making levels):
Each device shuld be essentially the same hardware as the current iopener with only minor variations in the manufacturing details. The devices each cover distinctly different segments within a single market (SOHO users and Linux/*BSD hackers) and could easly be sold at competitive and profitable prices.
The fact that Netpliance was able to offer the iopener at all indicates that it is only a matter of time before someone is offering these products to consumers. There is no good reason that Netpliance should be that someone.
For serious FP work the only reasonable choice is an Alpha: x86 just doesn't deliver very good FP performance. If you check out the Spec-95 scores (at http://www.spec.org/osg/cpu95/results/cfp95.html) for various systems you will find that, on SpecFP-95, Alphas out-perform almost all other systems by about 2:1 (with the exception of PA-RISC where the difference is only about 3:2). While x86 systems have been able to scale up the integer performance of the architecture, the FP performance has lagged considerably, probably do to the stack oriented FP register design (although it is possible that neither Intel nor AMD have seen FP performance as a prime motivator for sales, and hence haven't felt the need to push it in their designs).
Aho, Sethi & Ullman is a very good book, though some people think it's a bit unreadable. My personal favorite is Allen I. Holub's Compiler Design in C from Prentice Hall, which is more readable than Aho etc. without sacrificing any of the details. Another text, and the one that got me interested in compilers, is James E. Hendrix's A Small C Compiler from M&T Books which is also very readale by much less technical than Holub or Aho etc.
Under any circumstance, your prof is blowing a little bit of smoke up your ass with the talk of Aho, Sethi & Ullman being outdated just because it was written in 1988. Most of what we know about compiler design was known by the mid-sixties, if not before (some FORTRAN compilers were Really Good (R), even if many early C and Pascal compilers sucked). Although there was some progress made in the seventies and eighties, the state of the art hasn't changed all that much in the past ten years.
how about a TLD for jane citizen who isn't part of the network infrastructure, a governmental agency, commercial entity, educational institution, non-profit organization, or a branch of the military? I'd like to see someithing like .pri or .pub for private individuals.
- if I modify a SIS stored file does that modify the copy other people see?
- are there really all that many in the way of duplicate files on most servers? If so, then there wasn't all that much 'information' being stored on the server.
This reminds me of Ted Nelson's Xanadu project, which stored files as lists of references to other files. It would seem to me that storing references to common sub-ranges of files would be a real win, while this just sounds like a cheap hack that may look good on a feature list.selling at a loss is called dumping and is illegal, at least in the U.S. The claims about Sony not making a profit off of the Playstation don't say that Sony was selling at a loss, just very near cost, and that any profit was made off the sale of games.
If MS were to try to sell a piece of hardware at a significant loss (and if the specs are at all reliable the loss would be quite significant) then they would be in even deeper doo-doo than they are already with the web-browser business.
By my (admittedly under informed) calculations, a box matching Nintendo Next's specs would cost almost $150 for parts alone, without the processor. Clever bookkeeping may be able to assign assembly, packaging and marketing costs away, but it would be very hard to cover up the raw parts costs.
What CPU is going to be out in the next six months (they'll need some lead time on the engineering) that will run at 1GHz and will cost less than $100? Something from Intel: unlikely due to cost. Something from IBM or Motorola: unlikely due to speed. Something from AMD: possible, but if they could do it I would think we would have heard more hype than this. I'm not inclined to believe a gaming rag, even on a good day, but this is just silly.
I ended up purchasing the LinuxPPC CD because I couldn't get the downloaded stuff to work right. (this was LinuxPPC 1999 with all the installer problems) but I'm not sorry to have spent a little cash for the CD. It arrived very rapidly and has been working flawlessly both at home (on an old PowerComputing PowerWave) and at work (on a PowerMac 7200/120). I plan to buy the 2000 version as well. (I was cheap and didn't get the free upgrade)
otherwise how would the linuxsucks site have been able to register the domain name?
These other folks were clearly just obnoxious, profiteering, domain squaters and deserved to be slapped down like the running dogs they are.
The main feature of CodeWarrior isn't the pretty IDE, it's the portable C++ GUI framework, PowerPlant. With PowerPlant you can write a single set of source code and recompile for any supported platform: MacOS, BeOS, Windows95/98/NT/etc, or X Windows. [X]Emacs doesn't do anything to address this, nor do most other IDEs, nor do any of the companies providing assorted compilers for different platforms.
Another feature of CodeWarrior, that was still in beta last time I owned a license on MacOS, was a Java compiler that produced binary executables instead of byte-code class files. While this may seem to defeat the primary purpose of Java (compile once, run anywhere) it produces much faster Java programs on the supported platforms.
A couple years ago I thought that having a version of CodeWarrior for Linux would be ideal. At the time I was doing almost all my programming on MacOS, with an eye toward porting to Windows. It seemed important, then, to be able to have a single code base (using PowerPlant) that would produce GUI executables for all target platforms.
Having an X version of PowerPlant would have been a big help, as would have been the ability to code in Java and compile straight to binary executables (something that could be done under CodeWarrior for MacOS)
These days, however, I'm doing pretty much all my coding for *nix/X Windows. With MacOS migrating to a Mach/BSD base I don't feel any need to write old-style MacOS versions of my code, and I figure that anyone using Windows can fend for themselves.
I think that CodeWarrior just missed their window of opportunity. A year or so ago there was a real need for good cross platform development tools. Now that it looks like the whole world will be running *nix in a few years, however, I don't think there is nearly as much interest (or need) for such things.
After thinking about it for a bit, I can see a Spiderman movie with a Raimi atmosphere being a Really Good Thing (TM). Raimi's TV productions have had just the right mix of light drama and outright zaniness for a Spidey movie. Besides, he can't do worse than Marvel group has been doing with their modern attempts at either animated or live action productions (anyone remember the Fantasitc Four movie that never even made it beyond the trailer? How about the Generation X abortion? And have you tried to watch any of the animated children's series? They are completely unwatchable: Even when you know what the plotlines are supposed to be, the cartoons are impossible to follow!)
DC has at least done reasonably well with their animated series, though the Batman movies went from acceptable to complete dreck. I'm very disappointed that Marvel has been squandering their best characters and plotlines in a frenzy of cheap marketing, obtuse media productions and rediculous revisionism.