Re:Digital cameras do need a portable hard drive
on
Terapin Mine Review
·
· Score: 2
What's the difference between a $100 SLR and a $5000 SLR? Not a whole hell of a lot, all the cost is in the optics.
Which is why I'm wondering why the person I was replying to was storing 4.4 megapixel images when the optics in his digital camera likely couldn't resolve to that level.
I could be talking out of my ass, but it seems to me that a $300 digital camera won't have great optics on it since all that cost is in the CCD and hardware.
Re:Digital cameras do need a portable hard drive
on
Terapin Mine Review
·
· Score: 2
I personally like to zoom way the hell in.
I suppose you have one of those $20k professional digital cameras where the optics don't limit your resolution no matter what kind of film you use. Yes sir, nothing like using the highest resolution film in a $75 fixed-focus camera.
OC-3, OC-12, OC-24, OC-48, and OC-192 can go 155.52Mbps, 622.08Mbps, 1.244Gbps, 2.488Gbps, and 9.952Gbps, respectively. So, there I shown you 5 links that can go faster than 12Mbps and further then a few kilometers, and all are serial.
No, you've shown me 5 optical links that span more than a few kilometres and are serial, which really isn't much at all. Single-mode fiber can go hundreds of kilometers without issue. Even telco-grade comms are limited though: A 100m coaxial DS3 (44.7Mbps) run is considered "long haul" by its termination equipment, and I believe the max. length for this type of connection is 135m. And that's on two coaxial lines!
100BaseTX can do 100m at 100Mbps using two pair in a 4-pair UTP cable; We're not seeing more links like this because the termination electronics are costly. Your endpoint baluns don't exist in USB and IIRC the USB protocol is quite a bit simpler than 802.3 which helps keep the silicon cost down. Using ethernet is quite popular in industrial segments: power another pair and have the fourth pair for 4-20mA analog signalling. A web search on ModBUS/TCP shows where industry is heading.
In short, thanks for playing. The article is about consumer-grade PC communications, not long-haul telco comms, and I stand behind my original argument in that context.
if quantum events down on that level start screwing with the punch holes after a while, and u gradually lose tiny bits (pun intended) of data after a while, then it won't replace optical data storage
This happens on a daily basis with disk, tape and CD. The data is not written exactly as it comes; it's interleaved and written with wads of error correction codes attached. That is what makes your CD-ROM or hard drive or tape seem perfect.
part from their products being way too expensive, USB is not a viable technology. As opposed to some serial standards which can go for kilometers, USB is limited to mere meters.
Show me a 12Mbps link that can go kilometres. It's all a speed/distance tradeoff.
Hell RS232 can only go a handful of meters as well. It's the differential standards that go the distance (RS422, RS485, etc.) -- USB is also differential but it's also much much higher speed.
One baby step closer to getting a 2D/3D drafting program to Linux like AutoDesk Mechanical Desktop and Inventor.
OrCAD for Linux would be great, although I've almost got SDT/PCB386 working under DOSEmu and VMWare. 800x600 with VMWare, 1024x768 for DOSEmu.:-)
Possibly the only thing holding my office at Windows is MPLAB ICE 2000. The damn thing won't install under WINE but works great under VMWare or Win4Lin. Office is licked now that StarOffice is here and Mozilla or Konq does all we need for web browsing.
Was to move all my shit into a 200 sq. foot office at the back of an ISP I work at. I take a hit on pay but I don't pay for the phone, space or 'net connection.
It actually works out a lot better than moving everything to the basement or something. I can have the lights where I want, have the music where I need it (or have total silence) and the house looks like a house, not a geek compound.
Actually there is one exception: I have the home-rolled DSL coming into the basement where the firewall has a wireless card. Her/the kids' computer is in the kitchen and when I bring my notebook home I can work wherever.
The only thing that blows about it all is that it's not exactly handy to scoot over here to do work on something when I am inspired at 3am.:-)
Since so few BIOS functions are actually used once the operating system gets into place, it's becoming less and less of a concern to get things perfect.
This is mostly true, but one part which every OS needs is correct ACPI information, which many BIOSs do not have.
Paradox? Hardly. No unix virus will exist since you can't infect binaries you don't have write permission for (i.e. all of them). And for most typical settings the user won't have access to a compiler to create their own binaries. Perl can be forced to run in taint mode. You can mount the/home directory as noexec, preventing any scripting at all. No viruses.
Trojans? Sure, but I can limit their damage as well. I can alias rm to prompt if more than 'x' number of files match the glob. Or to confirm any deletion. Hell, I can alias rm to move to a trash folder/chown admin instead of erasing. IIRC many nasty featues in DCOP require user confirmation or can be turned off altogether. Trojans are more of a problem than viruses, but are not nearly the problem that they are on Windows.
And most important of all, your/home is the exact thing that is backed up nightly!
Paradox? Bullshit. A full ten secods of thought would have prevented you from posting that comment.
I bet if the 'market share' of Windows and Linux was reversed, there'd be Linux viruses taking advantage of every root expliot available.
This happens already; people are already taking advantage of every remote (and local) root exploit available. There are many examples of poor programming on BUGTRAQ et al showing that linux applications can be just as poorly written. Marketshare has absolutely nothing to do with viruses.
You have a team of mechanics, and for the last 20 years all they have serviced, as well as driven themselves are Ford automobiles. Now, your boss tells them to jump right in and service Chevrolet autos too.
So now what you're saying is that it really won't be that big a deal. The devil's in the details, of course, but according to that analogy it shouldn't be all that hard.
Red Hat's additions make incompatibilities with other Linux distributions, and the company seems to follow an "embrace and extend" pattern like Microsoft does that forces companies to use Red Hat if they want the best compatibility with Red Hat...
+9 Right On the Money, Bay-bee!
I have hated RedHat distributions for three reasons: completely fucked up configuration systems, that abomination.RPM system and last but definately not least: proprietary kernel patches.
If the features were at all relevant to the general Linux user they would have been incorporated into the standard Linux kernel by now. Give me a pure kernel, a pure packaging and call it Slackware.
They have about the same processing power, which is why a lot of games get ported. But the SNES is a 16 bit version of the 6502 (Nintendo tried for NES back compatibility, but it didn't work out), whereas the GBA is a 32 bit ARM processor.
Let me get this straight. A 10MHz steroid-addled 6502 could keep pace with a 32-bit ARM processor? Did I read that right? Or does the GBA really have a lot more processing power that isn't being tapped?
Yeah, let's transfer megabytes of data over a searial cable. Sure, that'll work!
Where are you transfering that amount of data? I would see maybe 100 bytes to 500 bytes for a single request or answer.
So let's use easy numbers: 1k per transaction. at 115200 baud that is 11 transactions per second, assuming 10 bits per character (8 data + 1 start and stop bit). The database shouldn't be a bottleneck there but serial cards these days can do 512kbaud easily (51 transactions/sec). Remember this is at 1k per transaction. A more realistic number would be around 200 bytes per transaction -- including packet framing and checksumming -- at which point 115kbaud gets you 57 transactions/sec and 512kbaud hits ~260 transactions/sec. If you've got requests for numbers exceeding that I think you can probably afford bigger and badder security hardware.
Hell, sync serial ports can do T1 speeds and faster without breaking a sweat. With a pair of v.35 cards your systems can now handle 1280 1k transactions/sec, or ~6550 transactions/sec for 200 byte transactions. (Remember that v.35 can do 10MBps at 90m cable lengths.)
Christ the ideas people write off without a shred of thought...
I hope you're trolling, because your question is unbelievably dense. You don't put the security on the live machine, you do it on the protected box. You're not out to eliminate credit card theft, but rather to limit it greatly. Who cares if you can get 1 or 10 or 100 card numbers? It's the "hax0r steals 10k card numbers from insecure site!" headlines you're trying to avoid, and that's exactly what this system will help prevent. It slows down the rate of card numbers you can get and anything excessive is easily alarmed.
Thieves don't often go for difficult targets, they want the fast big payoff. This system prevents that, although I think they're being excessive with the serial ports with snipped leads. Two NICs in the live host and one in the protected system going through a VLAN-capable switch or a router with the bandwidth on the connection set to 8kbps or something is just as good, IMO.
Isn't the user needing to reenter details to reauthorise transactions more of a feature than a bug?
Not with the general public. "XYZ Co. doesn't make me have to reenter all this credit number stuff like you do and I've never had my number stolen with them!" Convenience is the game.
ISTR that if you have proven that you took reasonable measures (co-host) to protect the credit card data and you were still 0wned and had it stolen that that would be enough to deflect loss of merchant status / lawsuits, although IANAL and I've never been in that position.
Yeah, until the fuxing disk drops data. I don't trust floppies for anything these days. Hell you could use a Zip (new ones, not the old click 'o death ones) or even cheap-ass 8MB CF which'll work for a year or so without flaking out.
I like the idea of having the data stored to disk but NOT floppies. Even the best quality floppies are shit these days.
I have yet to see it completely crash to the point where I need to reload it (I've had it for a little over a year), which is something I can't say for Palm.
That's bullshit.
I've owned 2 Palms in the past 5 years -- I had a Pro and then an early Vx (the case says V but it's got 8M and has not been modded) -- anyway in the past 5 years I have never had to reinstall, ever. That includes screwing around with TRG Flash, developing applications for it and daily use and abuse.
As a matter of fact, I don't think I've heard of anyone having to reinstall all their apps on their Palm (any version). Now if you let it go dead then sure but I haven't managed to do that, even with my shitty memory.
My copy tuned up yesterday and I started reading it in the bath this morning
The reviews and the sample pages didn't get into it -- how is it for depth? As an example, its chapter on USB interfacing seems rather... thin. How does it fare? Are there schematics? Is it an "Embedded interfacing for Dummies" type of book or is there actually some meat to it?
Well, your PDF breaks Acrobat Reader 5.01 - page down from the title page gives "There was an error processing a page. The page contents object has the wrong type." The table of contents doesn't seem to start until page 5.
Weird... It works beautifully on Acrobat Reader 4 for Linux.
The drives are configured as software RAID 5 and ext3.
Ext3 on a production server? You're braver than I... it's still an experimental FS (from 2.4.18 make config):
Ext3 journalling file system support (EXPERIMENTAL)
IDE will work fine for most small offices and you've got the "no slave devices" right for RAID. My personal preference for small offices is two large IDE disks in RAID1 with a tape backup.
Linear regulators like the 78xx/79xx and LM317 and co. are all fine and dandy but you'll be whoring a lot of power out as heat. I'm also not sure that they have the regulation required to meet the various power quality specs that motherboards require. Go switchmode and get your efficiencies (and step responses) up.
Or, do what everyone in the telco business does and buy DC AT/ATX power supplies. -48VDC in, regular switchmode supply for all your motherboard needs. And no UPS required.
Slashdot Mirror
What's the difference between a $100 SLR and a $5000 SLR? Not a whole hell of a lot, all the cost is in the optics.
Which is why I'm wondering why the person I was replying to was storing 4.4 megapixel images when the optics in his digital camera likely couldn't resolve to that level.
I could be talking out of my ass, but it seems to me that a $300 digital camera won't have great optics on it since all that cost is in the CCD and hardware.
I personally like to zoom way the hell in.
I suppose you have one of those $20k professional digital cameras where the optics don't limit your resolution no matter what kind of film you use. Yes sir, nothing like using the highest resolution film in a $75 fixed-focus camera.
OC-3, OC-12, OC-24, OC-48, and OC-192 can go 155.52Mbps, 622.08Mbps, 1.244Gbps, 2.488Gbps, and 9.952Gbps, respectively. So, there I shown you 5 links that can go faster than 12Mbps and further then a few kilometers, and all are serial.
No, you've shown me 5 optical links that span more than a few kilometres and are serial, which really isn't much at all. Single-mode fiber can go hundreds of kilometers without issue. Even telco-grade comms are limited though: A 100m coaxial DS3 (44.7Mbps) run is considered "long haul" by its termination equipment, and I believe the max. length for this type of connection is 135m. And that's on two coaxial lines!
100BaseTX can do 100m at 100Mbps using two pair in a 4-pair UTP cable; We're not seeing more links like this because the termination electronics are costly. Your endpoint baluns don't exist in USB and IIRC the USB protocol is quite a bit simpler than 802.3 which helps keep the silicon cost down. Using ethernet is quite popular in industrial segments: power another pair and have the fourth pair for 4-20mA analog signalling. A web search on ModBUS/TCP shows where industry is heading.
In short, thanks for playing. The article is about consumer-grade PC communications, not long-haul telco comms, and I stand behind my original argument in that context.
if quantum events down on that level start screwing with the punch holes after a while, and u gradually lose tiny bits (pun intended) of data after a while, then it won't replace optical data storage
This happens on a daily basis with disk, tape and CD. The data is not written exactly as it comes; it's interleaved and written with wads of error correction codes attached. That is what makes your CD-ROM or hard drive or tape seem perfect.
I agree. Email works too, but email isn't centrally archived and isn't searchable by everyone (except for the email admin, ideally).
So set up a mailing list instead. It's still email, and it's archived, threaded and searchable.
part from their products being way too expensive, USB is not a viable technology. As opposed to some serial standards which can go for kilometers, USB is limited to mere meters.
Show me a 12Mbps link that can go kilometres. It's all a speed/distance tradeoff.
Hell RS232 can only go a handful of meters as well. It's the differential standards that go the distance (RS422, RS485, etc.) -- USB is also differential but it's also much much higher speed.
One baby step closer to getting a 2D/3D drafting program to Linux like AutoDesk Mechanical Desktop and Inventor.
OrCAD for Linux would be great, although I've almost got SDT/PCB386 working under DOSEmu and VMWare. 800x600 with VMWare, 1024x768 for DOSEmu. :-)
Possibly the only thing holding my office at Windows is MPLAB ICE 2000. The damn thing won't install under WINE but works great under VMWare or Win4Lin. Office is licked now that StarOffice is here and Mozilla or Konq does all we need for web browsing.
Was to move all my shit into a 200 sq. foot office at the back of an ISP I work at. I take a hit on pay but I don't pay for the phone, space or 'net connection.
It actually works out a lot better than moving everything to the basement or something. I can have the lights where I want, have the music where I need it (or have total silence) and the house looks like a house, not a geek compound.
Actually there is one exception: I have the home-rolled DSL coming into the basement where the firewall has a wireless card. Her/the kids' computer is in the kitchen and when I bring my notebook home I can work wherever.
The only thing that blows about it all is that it's not exactly handy to scoot over here to do work on something when I am inspired at 3am. :-)
Since so few BIOS functions are actually used once the operating system gets into place, it's becoming less and less of a concern to get things perfect.
This is mostly true, but one part which every OS needs is correct ACPI information, which many BIOSs do not have.
And that's the real paradox of Unix security
Paradox? Hardly. No unix virus will exist since you can't infect binaries you don't have write permission for (i.e. all of them). And for most typical settings the user won't have access to a compiler to create their own binaries. Perl can be forced to run in taint mode. You can mount the /home directory as noexec, preventing any scripting at all. No viruses.
Trojans? Sure, but I can limit their damage as well. I can alias rm to prompt if more than 'x' number of files match the glob. Or to confirm any deletion. Hell, I can alias rm to move to a trash folder/chown admin instead of erasing. IIRC many nasty featues in DCOP require user confirmation or can be turned off altogether. Trojans are more of a problem than viruses, but are not nearly the problem that they are on Windows.
And most important of all, your /home is the exact thing that is backed up nightly!
Paradox? Bullshit. A full ten secods of thought would have prevented you from posting that comment.
I bet if the 'market share' of Windows and Linux was reversed, there'd be Linux viruses taking advantage of every root expliot available.
This happens already; people are already taking advantage of every remote (and local) root exploit available. There are many examples of poor programming on BUGTRAQ et al showing that linux applications can be just as poorly written. Marketshare has absolutely nothing to do with viruses.
You have a team of mechanics, and for the last 20 years all they have serviced, as well as driven themselves are Ford automobiles. Now, your boss tells them to jump right in and service Chevrolet autos too.
So now what you're saying is that it really won't be that big a deal. The devil's in the details, of course, but according to that analogy it shouldn't be all that hard.
Red Hat's additions make incompatibilities with other Linux distributions, and the company seems to follow an "embrace and extend" pattern like Microsoft does that forces companies to use Red Hat if they want the best compatibility with Red Hat...
+9 Right On the Money, Bay-bee!
I have hated RedHat distributions for three reasons: completely fucked up configuration systems, that abomination .RPM system and last but definately not least: proprietary kernel patches.
If the features were at all relevant to the general Linux user they would have been incorporated into the standard Linux kernel by now. Give me a pure kernel, a pure packaging and call it Slackware.
They have about the same processing power, which is why a lot of games get ported. But the SNES is a 16 bit version of the 6502 (Nintendo tried for NES back compatibility, but it didn't work out), whereas the GBA is a 32 bit ARM processor.
Let me get this straight. A 10MHz steroid-addled 6502 could keep pace with a 32-bit ARM processor? Did I read that right? Or does the GBA really have a lot more processing power that isn't being tapped?
Yeah, let's transfer megabytes of data over a searial cable. Sure, that'll work!
Where are you transfering that amount of data? I would see maybe 100 bytes to 500 bytes for a single request or answer.
So let's use easy numbers: 1k per transaction. at 115200 baud that is 11 transactions per second, assuming 10 bits per character (8 data + 1 start and stop bit). The database shouldn't be a bottleneck there but serial cards these days can do 512kbaud easily (51 transactions/sec). Remember this is at 1k per transaction. A more realistic number would be around 200 bytes per transaction -- including packet framing and checksumming -- at which point 115kbaud gets you 57 transactions/sec and 512kbaud hits ~260 transactions/sec. If you've got requests for numbers exceeding that I think you can probably afford bigger and badder security hardware.
Hell, sync serial ports can do T1 speeds and faster without breaking a sweat. With a pair of v.35 cards your systems can now handle 1280 1k transactions/sec, or ~6550 transactions/sec for 200 byte transactions. (Remember that v.35 can do 10MBps at 90m cable lengths.)
Christ the ideas people write off without a shred of thought...
Where's the security in that ?
I hope you're trolling, because your question is unbelievably dense. You don't put the security on the live machine, you do it on the protected box. You're not out to eliminate credit card theft, but rather to limit it greatly. Who cares if you can get 1 or 10 or 100 card numbers? It's the "hax0r steals 10k card numbers from insecure site!" headlines you're trying to avoid, and that's exactly what this system will help prevent. It slows down the rate of card numbers you can get and anything excessive is easily alarmed.
Thieves don't often go for difficult targets, they want the fast big payoff. This system prevents that, although I think they're being excessive with the serial ports with snipped leads. Two NICs in the live host and one in the protected system going through a VLAN-capable switch or a router with the bandwidth on the connection set to 8kbps or something is just as good, IMO.
Isn't the user needing to reenter details to reauthorise transactions more of a feature than a bug?
Not with the general public. "XYZ Co. doesn't make me have to reenter all this credit number stuff like you do and I've never had my number stolen with them!" Convenience is the game.
ISTR that if you have proven that you took reasonable measures (co-host) to protect the credit card data and you were still 0wned and had it stolen that that would be enough to deflect loss of merchant status / lawsuits, although IANAL and I've never been in that position.
I think this is a job for the old floppy disk.
Yeah, until the fuxing disk drops data. I don't trust floppies for anything these days. Hell you could use a Zip (new ones, not the old click 'o death ones) or even cheap-ass 8MB CF which'll work for a year or so without flaking out.
I like the idea of having the data stored to disk but NOT floppies. Even the best quality floppies are shit these days.
A lot of people don't know about the "watch" command
Indeed. I always did while true ; do clear ; date ; sleep 3 ; done when I needed to keep an eye on something.
I never knew about the history command, although I use ^R in bash a whole lot.
du -h --max-depth 1 /home | sort -rn is always a favourite of mine; add a | head -10 to grab just the top 10 (or whatever you prefer).
I have yet to see it completely crash to the point where I need to reload it (I've had it for a little over a year), which is something I can't say for Palm.
That's bullshit.
I've owned 2 Palms in the past 5 years -- I had a Pro and then an early Vx (the case says V but it's got 8M and has not been modded) -- anyway in the past 5 years I have never had to reinstall, ever. That includes screwing around with TRG Flash, developing applications for it and daily use and abuse.
As a matter of fact, I don't think I've heard of anyone having to reinstall all their apps on their Palm (any version). Now if you let it go dead then sure but I haven't managed to do that, even with my shitty memory.
My copy tuned up yesterday and I started reading it in the bath this morning
The reviews and the sample pages didn't get into it -- how is it for depth? As an example, its chapter on USB interfacing seems rather... thin. How does it fare? Are there schematics? Is it an "Embedded interfacing for Dummies" type of book or is there actually some meat to it?
Well, your PDF breaks Acrobat Reader 5.01 - page down from the title page gives "There was an error processing a page. The page contents object has the wrong type." The table of contents doesn't seem to start until page 5.
Weird... It works beautifully on Acrobat Reader 4 for Linux.
The drives are configured as software RAID 5 and ext3.
Ext3 on a production server? You're braver than I... it's still an experimental FS (from 2.4.18 make config):
IDE will work fine for most small offices and you've got the "no slave devices" right for RAID. My personal preference for small offices is two large IDE disks in RAID1 with a tape backup.
Linear regulators like the 78xx/79xx and LM317 and co. are all fine and dandy but you'll be whoring a lot of power out as heat. I'm also not sure that they have the regulation required to meet the various power quality specs that motherboards require. Go switchmode and get your efficiencies (and step responses) up.
Or, do what everyone in the telco business does and buy DC AT/ATX power supplies. -48VDC in, regular switchmode supply for all your motherboard needs. And no UPS required.
You don't need to be paranoid when working with lethal voltages... but you do need an ounce of common sense. Work with one hand.
That's not even enough. 60Hz will capacitively couple quite nicely through an inch of rubber sole to concrete.
One hand helps a lot, but it doesn't mean you can go grabbing conductors. :-)