I agree that g4u doesn't sound like the best solution, at least for Linux installations. But even for occasional disk mirroring from a rescue floppy, one can do better than "nc" with little or no effort.
A technically better solution is probably to use multicast rsync, either on the raw partition, or on the mounted file system. Using it on a mounted file system has the advantage that it works on live file systems, can deal with different drive geometries, and doesn't waste any time copying free blocks that still contain data.
If you do use "nc", there are two things you should do first: (1) clear out any free data on the source partitions by "cat/dev/zero > junk; rm junk" (this will improve compression), and (2) use gzip, as in "gzip/dev/hda".
After being a long-time Solaris user, the steep price of the Tadpole was the last straw when I needed a UNIX laptop--that's what made me switch to Linux, and I have never looked back.
Good development tools? You've got to be kidding. The GNU project, gcc, gdb, command line tools, was a huge success on SunOS/Solaris early on, exactly because it filled the many gaps in Sun's operating system and development tools.
I'm not condoning copyright infringement, but this story is spin. Just because the same kind of greedy forces that are controlling copyright in the US are starting to get powerful in China doesn't mean that the Chinese are "realizing" anything. And the totalitarian methods of the Chinese state make this even uglier than the web of corporate and political power in the US.
Both China and the US should find copyright and fair use rules that benefit the people as a whole. Both countries, however, seem to choose copyright rules and laws that mostly benefit a few powerful minorities.
It's expensive because marketing believes that those kinds of devices are being sold to upper management types with lost of disposable cash. Of course, a lot of Logitech's products seem similarly overpriced.
Competition will drive this down to reasonable levels pretty soon.
They probably end up getting a lot more phone calls because they don't let people write Linux drivers. If companies like Logitech simply put the Windows driver source and/or some of their engineering documents on the web, Linux users wouldn't bother to call them.
Acroread and xpdf are great for presentations, and they run on Linux. Both have full screen modes and have pretty much the right bindings to be used with a remote control like this.
If you must use something like PowerPoint, StarOffice is, of course, another choice.
It's good to see that some Bluetooth devices are coming out, but for now, there are cheaper and simpler solutions.
For under $80, you can get the Keyspan Presentation Remote, which is smaller, comes with a carrying case, includes a laser pointer, and looks just like a USB mouse from the PC side (no drivers). And, yes, it works with Linux.
Another choice is the Gyration wireless mice. But their receiver is a bulky box and requires a wall wart. And if you want to be able to use standard AAA batteries, you have to pay $100 extra for the "Pro" version.
I really don't care very much about what kind of protocol I use to talk to the IM server. What bugs me is that I need to sign up with half a dozen different services in order to have a good chance of reaching most people.
What we really need is interoperability at the back-end--AOL IM servers need to talk to MSN IM and to IRC. Maybe standardized protocols would help with that a little (the AOL server could pretend to be a client for MSN), but I suspect lack of connectivity is more of a business thing.
Oh, sorry, I didn't see the "three interface" requirement. To add two more interfaces with the least amount of hassle, I'd still go with the CaseOutlet case and use USB Ethernet adapters. They are under $20 and USB powered. Some of the nicer ones (a little more expensive) are barely larger than a plug.
That would not only give you serial console management, but also very fast Linux boots.
How much would a similar 3-iface Mini-ITX system cost?
Pretty much the same amount: motherboard, CPU, case, PCI riser, screws, cables, and brick power supply from CaseOutlet.com costs a little under $200. Add to that a $20 CF-to-IDE adapter and the DWL-520. If you want to use lower-cost 5 1/4" drives, you can get a slightly larger case for slightly more money.
Another choice is the WalMart Linux PC, which appears to be using same motherboard, but for $228 also gives you a minitower, a CD-ROM drive, and a 10G disk, and perhaps more fans than you like.
I noted that VIA's suggested power supplies either had fans, except for that TK fanless ATX supply (which costs $150). I'm using a wall brick for my net4501.
There are cheap, wall-brick power supplies for these boards; the CaseOutlet cases come with one.
Mini-ITX system *must* include the net-booting eth card and other "accessories", the power supply that probably has moving parts, and the additional heat output
The onboard Ethernet net-boots; there isn't much else that you need, except maybe a $20 CF-to-IDE adapter.
I expect the Mini-ITX stuff (esp. w/ enough gear to do what the net4501 does) draws a lot more power than the 0.06A at 120V during normal (not idle) operation.
The 533MHz supposedly consumes 1W, and the 800MHz 5W (I haven't measured it).
The Mini-ITX stuff is very new, using very new technology
I dunno--looks like pretty run-of-the-mill PC technology to me.
with three eth ifaces, two 'pci' slots (one mini-pci, one 3.3v pci), and two serial ports
Well, no, but it comes with single or dual regular PCI slots (depending on riser), dual USB, Ethernet, VGA, S-video, Audio, PS/2, and serial, probably a more useful collection for many applications.
Well, since "WEP" doesn't actually provide anything near "wired equivalent privacy", it makes sense to change what the acronym stands for. "Wimpy Encryption Protocol" might be a better choice, though.
The Soekris motherboard costs $196, for a 133MHz 486. A Mini ITX motheboard with processor costs around $100, with your choice of 533MHz or 800MHz Pentium-compatible processor, plus a $20 CF-to-IDE adapter.
The Soekris has a number of advantages, primarily that it's smaller and that it is happy with just 5V or 7-20V DC. But you pay a premium for those features, and you sacrifice functionality. The Mini ITX gives you a standard PCI slot, many more I/O options, much better performance, and more I/O ports.
I think, given its functionality, the "value" of something like the Soekris 486 boards really "should be" around $50 these days, and that's what it probably would be if it really were manufactured in huge quantities. Does anybody know of a low-cost 486 PC104 board like that?
Forth is great when you need an interactive language and have very limited amounts of memory available. Something like a boot ROM or a tiny embedded system. You can squeeze a full, interactive system into a couple of kilobytes.
But that is probably its main virtue. Once you move development "off board" and use a batch compiler, you might as well use a compiler (C, Modula-2, etc.).
Forth and Postscript share that both use RPN (reverse polish notation), and as a consequence have some similarities for stack manipulation. So do HP calculators and the Java virtual machine. Beyond that, however, the languages couldn't be more different. Forth is a low-level language with no runtime safety and no dynamic typing. Postscript is a dynamically typed language with garbage collection. Except for the use of RPN notation, Postscript is really closer to languages like Lisp.
As far as I can tell, the big change with Quartz Extreme is that it uses the graphics hardware for compositing operations; but I don't think it uses it for accelerating all 2D drawing operations. That makes it great for window moving and image operations, but it may not help a lot with other kinds of graphics.
A high-performance, accelerated, anti-aliased graphics subsystem still seems some way off for both OS X and Linux/X11: OS X has high quality graphics throughout but is somewhat slow and heavy, while Linux/X11 is much smaller and can be very fast for the common cases, but anti-aliasing and transparency are only gradually becoming widely available. I expect the two graphics systems to end up being pretty similar in terms of functionality and performance in a couple of years: client/server 2D graphics systems with full Postscript-like imaging and hardware acceleration.
The only things that matter are the codec specifications, with the streaming protocols a distant second. And those only matter because there is a lot of RealAudio and RealVideo out there on the web, not because there aren't good open source alternatives already.
And, guess what, the codecs for Real's audio and video formats are just what they aren't releasing in source form.
All we are getting with Helix DNA is a lot of useless infrastructure code--no meat. Thanks, but no thanks. This is a useless marketing ploy.
It is hard to see why OS X should be far behind Linux when it comes to server-side speed: OS X is based on a mature kernel with reasonably mature file systems.
Where OS X seems behind Linux in terms of performance is memory footprint and graphics speed. In Moshe's server benchmarks, OS X used a lot more memory. Also, take a look at the memory footprint of the OS X display server and the OS X GUI applications--they are usually several times as large as comparable X11 functionality. And the raw graphics performance of the OS X display server is behind X11 in my experience (but do your own measurements if you don't believe me).
The Mac desktop market is significantly larger than the linux desktop market, so it's not a marketshare issue.
Can you back up that claim with some facts? And even if it's true, what fractions of the two markets are actually "hacker" types who would be interested in such a product?
Slashdot Mirror
A technically better solution is probably to use multicast rsync, either on the raw partition, or on the mounted file system. Using it on a mounted file system has the advantage that it works on live file systems, can deal with different drive geometries, and doesn't waste any time copying free blocks that still contain data.
If you do use "nc", there are two things you should do first: (1) clear out any free data on the source partitions by "cat /dev/zero > junk; rm junk" (this will improve compression), and (2) use gzip, as in "gzip /dev/hda".
After being a long-time Solaris user, the steep price of the Tadpole was the last straw when I needed a UNIX laptop--that's what made me switch to Linux, and I have never looked back.
Good development tools? You've got to be kidding. The GNU project, gcc, gdb, command line tools, was a huge success on SunOS/Solaris early on, exactly because it filled the many gaps in Sun's operating system and development tools.
Both China and the US should find copyright and fair use rules that benefit the people as a whole. Both countries, however, seem to choose copyright rules and laws that mostly benefit a few powerful minorities.
Competition will drive this down to reasonable levels pretty soon.
They probably end up getting a lot more phone calls because they don't let people write Linux drivers. If companies like Logitech simply put the Windows driver source and/or some of their engineering documents on the web, Linux users wouldn't bother to call them.
If you must use something like PowerPoint, StarOffice is, of course, another choice.
For under $80, you can get the Keyspan Presentation Remote, which is smaller, comes with a carrying case, includes a laser pointer, and looks just like a USB mouse from the PC side (no drivers). And, yes, it works with Linux.
Another choice is the Gyration wireless mice. But their receiver is a bulky box and requires a wall wart. And if you want to be able to use standard AAA batteries, you have to pay $100 extra for the "Pro" version.
Here:
l
http://www.caseoutlet.com/NWPc/2677/itx2677.htm
Get the Eden 533MHz version for a fanless motherboard.
What we really need is interoperability at the back-end--AOL IM servers need to talk to MSN IM and to IRC. Maybe standardized protocols would help with that a little (the AOL server could pretend to be a client for MSN), but I suspect lack of connectivity is more of a business thing.
Oh, sorry, I didn't see the "three interface" requirement. To add two more interfaces with the least amount of hassle, I'd still go with the CaseOutlet case and use USB Ethernet adapters. They are under $20 and USB powered. Some of the nicer ones (a little more expensive) are barely larger than a plug.
http://www.viavpsd.com/product/Download.jsp?mother boardId=21
Note also the upcoming EPIA-M motherboard, which features USB2, FireWire, and 933MHz clock speed.
As for serial consoles, it appears that the Linux Bios has been ported to it:
http://www.trustytech.com/TMBM-MINI-ITX.htm
That would not only give you serial console management, but also very fast Linux boots.
How much would a similar 3-iface Mini-ITX system cost?
Pretty much the same amount: motherboard, CPU, case, PCI riser, screws, cables, and brick power supply from CaseOutlet.com costs a little under $200. Add to that a $20 CF-to-IDE adapter and the DWL-520. If you want to use lower-cost 5 1/4" drives, you can get a slightly larger case for slightly more money.
Another choice is the WalMart Linux PC, which appears to be using same motherboard, but for $228 also gives you a minitower, a CD-ROM drive, and a 10G disk, and perhaps more fans than you like.
There are cheap, wall-brick power supplies for these boards; the CaseOutlet cases come with one.
Mini-ITX system *must* include the net-booting eth card and other "accessories", the power supply that probably has moving parts, and the additional heat output
The onboard Ethernet net-boots; there isn't much else that you need, except maybe a $20 CF-to-IDE adapter.
The 533MHz supposedly consumes 1W, and the 800MHz 5W (I haven't measured it).
The Mini-ITX stuff is very new, using very new technology
I dunno--looks like pretty run-of-the-mill PC technology to me.
with three eth ifaces, two 'pci' slots (one mini-pci, one 3.3v pci), and two serial ports
Well, no, but it comes with single or dual regular PCI slots (depending on riser), dual USB, Ethernet, VGA, S-video, Audio, PS/2, and serial, probably a more useful collection for many applications.
No, it wouldn't. WEP is just broken--it can be listened into with one of a number of simple software downloads.
Well, since "WEP" doesn't actually provide anything near "wired equivalent privacy", it makes sense to change what the acronym stands for. "Wimpy Encryption Protocol" might be a better choice, though.
The Soekris has a number of advantages, primarily that it's smaller and that it is happy with just 5V or 7-20V DC. But you pay a premium for those features, and you sacrifice functionality. The Mini ITX gives you a standard PCI slot, many more I/O options, much better performance, and more I/O ports.
I think, given its functionality, the "value" of something like the Soekris 486 boards really "should be" around $50 these days, and that's what it probably would be if it really were manufactured in huge quantities. Does anybody know of a low-cost 486 PC104 board like that?
- Motherboard and case from CaseOutlet.com; you get a choice of 533MHz or 800MHz, for about $200. Motherboard and CPU alone are $100.
- CF-to-IDE adapter from various sources, for about $20
- 32MB CF card for $15 (512M for about $175)
- wireless PCI 802.11b adapter, for about $40
Total cost: $275.This takes a few minutes to put together. You get a choice between 12V or 120V power supply.
But that is probably its main virtue. Once you move development "off board" and use a batch compiler, you might as well use a compiler (C, Modula-2, etc.).
Forth and Postscript share that both use RPN (reverse polish notation), and as a consequence have some similarities for stack manipulation. So do HP calculators and the Java virtual machine. Beyond that, however, the languages couldn't be more different. Forth is a low-level language with no runtime safety and no dynamic typing. Postscript is a dynamically typed language with garbage collection. Except for the use of RPN notation, Postscript is really closer to languages like Lisp.
A high-performance, accelerated, anti-aliased graphics subsystem still seems some way off for both OS X and Linux/X11: OS X has high quality graphics throughout but is somewhat slow and heavy, while Linux/X11 is much smaller and can be very fast for the common cases, but anti-aliasing and transparency are only gradually becoming widely available. I expect the two graphics systems to end up being pretty similar in terms of functionality and performance in a couple of years: client/server 2D graphics systems with full Postscript-like imaging and hardware acceleration.
And, guess what, the codecs for Real's audio and video formats are just what they aren't releasing in source form.
All we are getting with Helix DNA is a lot of useless infrastructure code--no meat. Thanks, but no thanks. This is a useless marketing ploy.
Where OS X seems behind Linux in terms of performance is memory footprint and graphics speed. In Moshe's server benchmarks, OS X used a lot more memory. Also, take a look at the memory footprint of the OS X display server and the OS X GUI applications--they are usually several times as large as comparable X11 functionality. And the raw graphics performance of the OS X display server is behind X11 in my experience (but do your own measurements if you don't believe me).
Slashdotters did this a while ago :-)
Can you back up that claim with some facts? And even if it's true, what fractions of the two markets are actually "hacker" types who would be interested in such a product?