I am a fan of both OpenWRT and Angstrom (formerly OpenZaurus) and have on more than one occasion used OpneEmbedded to build packages and whole binary distributions for my Zauruses (Zaurii?). Having said that I don't think that OpenEmbedded is really applicable in this case. OE's big focus is on making cross compiling less painful and on offering a more fine grained approach to the creation of binary packages. Since Intel's UMPC platform doesn't need to be cross compiled for (it's plain x86) and since it has lots of storage space at it's disposable (a minimum of 4GB or so as opposed to 16MB or less on on some OE supported targets), the benefits of OE might be outweighed when compared to the number of available source packages from a standard desktop Linux distro like Ubuntu or Debian brings in "for free." Just my 2 cents.
You don't have a history of trolling so I'm giving you the benefit of the doubt: - No version of the Linux kernel has ever been released under the GPL 3 - AFAICT, none of the articles mention anything about Intel using a proprietary kernel - On the projects page of moblin.org they list under the "Kernel" sub-heading that they will be working on "Platform-specific kernel patches and device drivers." I don't see any reason not to take this at face value.
I would encourage you to RTFA (either moblin.org or the linuxdevices article) where it talks about Intel using Hildon (the UI framework from Maemo). To me that looks like the biggest reusable chunk of open source code that Nokia has turned out so far. Much of the rest of Nokia's stuff is either off-the-shelf (the kernel, packaging system (apt+dpkg)) or closed source (media codecs, DSP code) or rather trivial (I don't think writing your own application launcher sidebar is really going to cause significant fracturing of the Linux userbase). There is lots of code reuse if you look for it a little deeper than in a slashdot summary...
Unfortunately in an Internet world you need to stamp a return address on every packet you send out in the form of an IP address. Yes, but it only needs to be *your* IP address if you want a response back. IIRC, you could sit their and send out UDP packets with spoofed source IPs all day long to some friendly host. It's not really a commonly accepted practice, but there's no one stopping you from doing it...
-Mr. Lizard
Re:And yes, you can play Theora in WMP
on
Dell Linux Details
·
· Score: 1
Uhm, no. What he's saying is it already works like that in Ubuntu 7.04. I can actually vouch for the fact that by default the built-in media player is a hell of a lot more compatible than Windows Media Player. Sometimes I've found the need to *gasp*shock* relaunch the media player (totem) before the newly installed codecs work. flash was either installed by default or trivially easy (ie, no command line needed). I actually find it funny that so many ubuntu howtos include instructions to use the command line for this and that when most of these things are easily accomplished through the GUI.
Meh, that was *so* last year's issue. Full, reliable read/write support for NTFS is available in most major distros. I would be surprised if a few of the leading edge distros don't already enable it by default.
And, real documentation for how.doc and the new "open" xml format would be nice but I bet in most situations either google's "documents and spreadsheets" or openoffice does a perfectly reasonable job of converting it to.odt and for simple documents saving back to.doc. People like to talk about corner cases where the formatting ends up slightly off, but seem to conveniently forget that different versions of MS Office are just as bad.
It's not like you can't run 32-bit code on a 64-bit kernel. I'm sure that applications running in 32-bit mode on 64-bit "Vienna" will have some restrictions imposed on them (such as not being able to access DRMed content, and somesuch) but it would be downright foolish of them to just up and make everyone compile their programs for x86_64. Microsoft has already shown that they care about *software* backwards compatibility quite a bit more than *hardware* backwards compatibility. Anyways, I'm willing to go on record as betting that most well-written, API following, 32-bit binaries compiled for Windows XP will work just fine on 64-bit Vienna.
I dunno. Apple seems to be doing okay with that strategy, and they don't even include a real word processor (no, TextEdit doesn't count) free with the machine. Many consumers are just not terribly attached to their software. If you can show them something that works almost exactly the same as the stuff they already have, they'll grumble a bit and then learn to cope. What Dell really needs to use as a selling point is the number of screensavers in Linux. It's like 10 times more than what Windows or Mac comes with. heh.
I can't believe you missed, "Driving a car that's not nice enough for the area, at the 'wrong' time of night." My friend had a really great one though: "Getting rear-ended by a white woman while being Asian and Hispanic." The real problem was the cops in that area never had anything better to do than harass people.
Also, in case you didn't notice, it seems like a lot of adware is saying the exact same thing: "Your computer is infected with spyware! You need to buy this tool to clean it!" How do you intend to differentiate your message from the dozens of pop-ups the user is already seeing?
The biggest threat to security is often from within the corporation/organization itself. And there's a big difference between being able to walk by someone's desk and see the sticky note with the password on it versus climbing under their desk and putting a key-logger between the system and the keyboard. Think about the following two scenarios:
Scenario 1: Worker: What were you doing going through the drawers in my desk for while I was away? Cracker: Sorry. I was looking for a stapler.
Scenario 2: Worker: What were you doing crawling around under my desk, screwing with my computer? Cracker: Sorry. I was looking for a stapler.
See, one of these is activities is a little more dubious than the other. Also, you don't have to be a 1337 hax0r to be a threat to security. All you have to do is have access to a file/account/system you shouldn't.
OpenEmbedded has exactly what he wants. My Sharp Zaurus C-1000, running OpenZaurus (built from the OE build system) has: a real x server, pretty desktop icons, gaim, abiword, gqview gaim, sylpheed, some games, an ebook reader, gftp, firefox and some other programs. All this takes up ~90MB flash. Also, the system is fairly comfortable to use even with only 64MB of RAM. I did setup swap on an SD card but that only gets hit when firefox and something else are running at the same time. With 128MB of RAM and a leaner browser (galeon or epiphany maybe?) I don't see a reason to use swap. If I was interested in a lightweight, maintainable Linux distro for this project, I'd make sure that the OE devs got hooked up with a development system (or:gasp: even *hired* to put prioritize OLPC support).
Just my $.02
-Mr. Lizard
The problem is that you want an apples-to-apples comparison of apples and oranges. The primary reason for hard drive failure is failure of the mechanical moving parts. The primary reason for flash drive failure is destroying a cell by writing to it too many times. Also, your statement about it being "trivially obvious that defragging any kind of drive reduces its lifespan" isn't quite as trivially obvious as you think. A hard drive will almost certainly suffer a mechanical failure long before it's gone through its allotment of spare blocks. On a flash drive that's written to a lot, bad blocks cropping up will probably be the first thing to go wrong.
If you buy a flash drive, fill it with data, and then never write to it again, you can read all you want and it's minimum MTBF will be ~10 years (AFAIK, there's no reason they couldn't last longer, it's just that more testing needs to be done to prove that they will last longer).
Another problem in comparing hard drives and flash drives is based on what kind of environment they're subjected to. Flash drives are usually portable devices that live in pockets, and are subject to static shocks and being plugged/unplugged on a regular basis. hard drives for the most part live in computers where they're protected from the elements and aren't often disconnected, especially not with the power on. In your case, I'd be willing to bet that your flash drives are dying from a failure in the onboard controller (rather than individual cells dying). It might be interesting to purchase a small USB hard drive and compare how long it lasts when subjected to the same environment as your flash drives.
-Mr. Lizard
Why isn't there a *real* bluetooth watch?
on
Top 10 Geek Watches
·
· Score: 2, Insightful
And yes I've seen all the prototypes and the computer renderings. It still doesn't mean I can go to a store and buy one off a shelf. Here is what I want this mythical bluetooth watch to do:
-When I have an incoming call it should: display the caller ID, light up the backlight on the watch, and maybe flash a little LED to get my attention. The watch needs one phone specific button to clear the alert and let the call ring out in silence then get sent to voice mail. That's all it needs. It would also be nice to have a button that answers the call on my bluetooth headset, but that's not really necessary. It should prevent the phone from ringing or even vibrating for a couple seconds so I have time to look at the caller ID and make a decision about the call. -It needs to do something similar for a text message. Maybe just flash the LED a couple times and display something like "new message from Bob" with the backlight lit. If I can read the text message on the phone that would be cool, but not strictly necessary. -I want a PDA or smartphone to be able to use the watch as a way to alert me of an impending meeting/appointment. The watch should have an open API to let other bluetooth devices talk to it and *be controlled by it* in at least a basic way. -If I can look through my contact list on my phone through it and use it to place calls and have the audio routed to a bluetooth headset, that would be cool. If not I can live without that feature. -Be something like a normal watch size. I will not wear something the size of a brick on my arm. -Have a readable monochrome LCD display. The wathces that were meant for use with that spot / msndirect thing seemed to have this figured out. I don't really have a sense of how big the watches were, but they didn't seem too huge... -Have a backlight that stays on for however long I hold the backlight button. This is not complex but I've had too many watches where the backlight would only stay on for a couple seconds at a time no matter whether you held the backlight button or not. It drove me nuts to be constantly jabbing that stupid backlight button.
It needs to *NOT* have the following: -wifi -a color screen -a GSM/CDMA radio -a fast processor -It's own unique copy of my phone book / contacts / appointments that must be synchronized to other devices. It should not be editing these, just displaying them when they're relavent.
This shouldn't really be that expensive. Bluetooth headsets are down to the $25 range, so obviously the bluetooth hardware isn't that expensive. And digital watches are way under $20 so that can't be a huge issue. This kind of a feature set is the only reason I'd ever be tempted to buy a watch again, so the sooner the watch industry gets on this, the sooner it makes a potential sale.
I would also take Gentoo over RedHat/Fedora any day of the week.;) And I understand that some people don't like the way Debian does things. What it comes down to is that *for me* Debian is the right compromise between number of packages, frequency of updates (I use "unstable" on my desktop and "stable" on production servers) and freedom to set my machine up the way I want. Yes, Gentoo gives you slightly more package selection and flexibility in setting up some things. And for some people "the Debian way" is a deal breaker.
It seems like emerge has the ability to easily mix source and binary packages. What I think would be really nice to see is a repository of binary packages for most of the core of the operating system, with the ability to easily force some packages to be compiled from source. For example it would be neat to install one's entire system from binaries; kernel, X11, gnome, mplayer, firefox, etc, then go back and set an override somewhere so that mplayer gets rebuilt from source to take advantage of all the nifty features of your processor. I'm not even saying that *all* packages need to be available as binaries. Heck, you could do a Debian style popularity contest and just keep the most popular packages around as binaries. I've always lived by the "Don't optimize all your code, just optimize the 10% that's being run 90% of the time." In my case I'd like to recompile a couple packages to select a certain non-default feature or enable a certain optimization, but it's not worth devoting the resources (disk space, CPU time, disk I/O) to compile every program I use from source.
While I'm making my wishlist, a program to log which libraries/programs are using the most CPU time, would be cool too. "Oh, I downloaded the binary package for libfoo, and it's pretty CPU intensive and a lot of programs link against it. Maybe I'll recompile it with some optimizations turned on." And, yes I've heard of "top" but I want something that sits in the background and will print out a nice report every so often. Once someone does all that, I'll drop Debian in a heartbeat, but for now it's the OS that annoys me the least.
The new installer looks very slick and I'd love to give this a try. However, I really would prefer to use binary packages. I know that I could get a minor speed improvements compiling by hand, but I have fast enough hardware these days that I'm willing to trade some speed for not having to wait around while Gnome compiles. So, where are these binary packages you speak of? Gentoo.org doesn't seem to list any central repository of pre-compiled packages, and a couple searches on Google have only told me that: 1) Gentoo supports binary packages -and- 2) I should just compile from source because it's better
So, is there an unofficial place to download binary packages that is something like a recognized authority? Where do I get these fabled binary packages? I'm not trolling, I really would like to know.
Yeah, but it would also be stupid easy to just "lose" that flag while doing a recording. If they don't want the audio to sound like shit, this "flag" has to be outside the human hearing range. If it's outside our hearing range, then just use an analog filter made out of cheap radioshack parts to hack off the flag. Then record it into a device that you *do* have end to end control of, like a tape recorder or any computer made before the great DRM law of 2007. Keep these current generation computers around! They might be useful once DRM is built into the hardware at a firmware level (and required by law...)
Supposedly, eInk screens only use power when changing the image they display. If that's true the batteries really will last for however long it takes you to read 7,500 pages. That is as long as you don't take so long to read 7,500 pages that the self discharge rate of whatever type of batteries it uses becomes a factor...
actually booting off of even a "slow" flash memory device, like an older usb drive will be quite quick. Much faster than booting off of a CD-ROM and quite close to the speed of booting off of a hard drive. During a normal boot process you're loading a lot of smallish programs/files, and this plays to the advantages of flash media: no seek times. CD-ROMs have seek times in the tens of milliseconds (maybe even 100 ms for an older unit). Harddrives less than 10 ms these days. Flash media on the other hand is truly random access in the same way that DRAM is, in that there isn't any kind of "seeking" done.
Slashdot Mirror
I am a fan of both OpenWRT and Angstrom (formerly OpenZaurus) and have on more than one occasion used OpneEmbedded to build packages and whole binary distributions for my Zauruses (Zaurii?). Having said that I don't think that OpenEmbedded is really applicable in this case. OE's big focus is on making cross compiling less painful and on offering a more fine grained approach to the creation of binary packages. Since Intel's UMPC platform doesn't need to be cross compiled for (it's plain x86) and since it has lots of storage space at it's disposable (a minimum of 4GB or so as opposed to 16MB or less on on some OE supported targets), the benefits of OE might be outweighed when compared to the number of available source packages from a standard desktop Linux distro like Ubuntu or Debian brings in "for free." Just my 2 cents.
You don't have a history of trolling so I'm giving you the benefit of the doubt:
- No version of the Linux kernel has ever been released under the GPL 3
- AFAICT, none of the articles mention anything about Intel using a proprietary kernel
- On the projects page of moblin.org they list under the "Kernel" sub-heading that they will be working on "Platform-specific kernel patches and device drivers." I don't see any reason not to take this at face value.
For more info please RTFA. kthxbye.
I would encourage you to RTFA (either moblin.org or the linuxdevices article) where it talks about Intel using Hildon (the UI framework from Maemo). To me that looks like the biggest reusable chunk of open source code that Nokia has turned out so far. Much of the rest of Nokia's stuff is either off-the-shelf (the kernel, packaging system (apt+dpkg)) or closed source (media codecs, DSP code) or rather trivial (I don't think writing your own application launcher sidebar is really going to cause significant fracturing of the Linux userbase). There is lots of code reuse if you look for it a little deeper than in a slashdot summary...
-Mr. Lizard
Uhm, no. What he's saying is it already works like that in Ubuntu 7.04. I can actually vouch for the fact that by default the built-in media player is a hell of a lot more compatible than Windows Media Player. Sometimes I've found the need to *gasp*shock* relaunch the media player (totem) before the newly installed codecs work. flash was either installed by default or trivially easy (ie, no command line needed). I actually find it funny that so many ubuntu howtos include instructions to use the command line for this and that when most of these things are easily accomplished through the GUI.
Meh, that was *so* last year's issue. Full, reliable read/write support for NTFS is available in most major distros. I would be surprised if a few of the leading edge distros don't already enable it by default.
.doc and the new "open" xml format would be nice but I bet in most situations either google's "documents and spreadsheets" or openoffice does a perfectly reasonable job of converting it to .odt and for simple documents saving back to .doc. People like to talk about corner cases where the formatting ends up slightly off, but seem to conveniently forget that different versions of MS Office are just as bad.
And, real documentation for how
It's not like you can't run 32-bit code on a 64-bit kernel. I'm sure that applications running in 32-bit mode on 64-bit "Vienna" will have some restrictions imposed on them (such as not being able to access DRMed content, and somesuch) but it would be downright foolish of them to just up and make everyone compile their programs for x86_64. Microsoft has already shown that they care about *software* backwards compatibility quite a bit more than *hardware* backwards compatibility. Anyways, I'm willing to go on record as betting that most well-written, API following, 32-bit binaries compiled for Windows XP will work just fine on 64-bit Vienna.
-Mr. Lizard
I dunno. Apple seems to be doing okay with that strategy, and they don't even include a real word processor (no, TextEdit doesn't count) free with the machine. Many consumers are just not terribly attached to their software. If you can show them something that works almost exactly the same as the stuff they already have, they'll grumble a bit and then learn to cope. What Dell really needs to use as a selling point is the number of screensavers in Linux. It's like 10 times more than what Windows or Mac comes with. heh.
-Mr. Lizard
I can't believe you missed, "Driving a car that's not nice enough for the area, at the 'wrong' time of night." My friend had a really great one though: "Getting rear-ended by a white woman while being Asian and Hispanic." The real problem was the cops in that area never had anything better to do than harass people.
Yes.
Also, in case you didn't notice, it seems like a lot of adware is saying the exact same thing: "Your computer is infected with spyware! You need to buy this tool to clean it!" How do you intend to differentiate your message from the dozens of pop-ups the user is already seeing?
The biggest threat to security is often from within the corporation/organization itself. And there's a big difference between being able to walk by someone's desk and see the sticky note with the password on it versus climbing under their desk and putting a key-logger between the system and the keyboard. Think about the following two scenarios:
Scenario 1:
Worker: What were you doing going through the drawers in my desk for while I was away?
Cracker: Sorry. I was looking for a stapler.
Scenario 2:
Worker: What were you doing crawling around under my desk, screwing with my computer?
Cracker: Sorry. I was looking for a stapler.
See, one of these is activities is a little more dubious than the other. Also, you don't have to be a 1337 hax0r to be a threat to security. All you have to do is have access to a file/account/system you shouldn't.
OpenEmbedded has exactly what he wants. My Sharp Zaurus C-1000, running OpenZaurus (built from the OE build system) has: a real x server, pretty desktop icons, gaim, abiword, gqview gaim, sylpheed, some games, an ebook reader, gftp, firefox and some other programs. All this takes up ~90MB flash. Also, the system is fairly comfortable to use even with only 64MB of RAM. I did setup swap on an SD card but that only gets hit when firefox and something else are running at the same time. With 128MB of RAM and a leaner browser (galeon or epiphany maybe?) I don't see a reason to use swap. :gasp: even *hired* to put prioritize OLPC support).
Just my $.02
-Mr. Lizard
If I was interested in a lightweight, maintainable Linux distro for this project, I'd make sure that the OE devs got hooked up with a development system (or
The problem is that you want an apples-to-apples comparison of apples and oranges. The primary reason for hard drive failure is failure of the mechanical moving parts. The primary reason for flash drive failure is destroying a cell by writing to it too many times. Also, your statement about it being "trivially obvious that defragging any kind of drive reduces its lifespan" isn't quite as trivially obvious as you think. A hard drive will almost certainly suffer a mechanical failure long before it's gone through its allotment of spare blocks. On a flash drive that's written to a lot, bad blocks cropping up will probably be the first thing to go wrong.
If you buy a flash drive, fill it with data, and then never write to it again, you can read all you want and it's minimum MTBF will be ~10 years (AFAIK, there's no reason they couldn't last longer, it's just that more testing needs to be done to prove that they will last longer).
Another problem in comparing hard drives and flash drives is based on what kind of environment they're subjected to. Flash drives are usually portable devices that live in pockets, and are subject to static shocks and being plugged/unplugged on a regular basis. hard drives for the most part live in computers where they're protected from the elements and aren't often disconnected, especially not with the power on. In your case, I'd be willing to bet that your flash drives are dying from a failure in the onboard controller (rather than individual cells dying). It might be interesting to purchase a small USB hard drive and compare how long it lasts when subjected to the same environment as your flash drives.
-Mr. Lizard
And yes I've seen all the prototypes and the computer renderings. It still doesn't mean I can go to a store and buy one off a shelf. Here is what I want this mythical bluetooth watch to do:
-When I have an incoming call it should: display the caller ID, light up the backlight on the watch, and maybe flash a little LED to get my attention. The watch needs one phone specific button to clear the alert and let the call ring out in silence then get sent to voice mail. That's all it needs. It would also be nice to have a button that answers the call on my bluetooth headset, but that's not really necessary. It should prevent the phone from ringing or even vibrating for a couple seconds so I have time to look at the caller ID and make a decision about the call.
-It needs to do something similar for a text message. Maybe just flash the LED a couple times and display something like "new message from Bob" with the backlight lit. If I can read the text message on the phone that would be cool, but not strictly necessary.
-I want a PDA or smartphone to be able to use the watch as a way to alert me of an impending meeting/appointment. The watch should have an open API to let other bluetooth devices talk to it and *be controlled by it* in at least a basic way.
-If I can look through my contact list on my phone through it and use it to place calls and have the audio routed to a bluetooth headset, that would be cool. If not I can live without that feature.
-Be something like a normal watch size. I will not wear something the size of a brick on my arm.
-Have a readable monochrome LCD display. The wathces that were meant for use with that spot / msndirect thing seemed to have this figured out. I don't really have a sense of how big the watches were, but they didn't seem too huge...
-Have a backlight that stays on for however long I hold the backlight button. This is not complex but I've had too many watches where the backlight would only stay on for a couple seconds at a time no matter whether you held the backlight button or not. It drove me nuts to be constantly jabbing that stupid backlight button.
It needs to *NOT* have the following:
-wifi
-a color screen
-a GSM/CDMA radio
-a fast processor
-It's own unique copy of my phone book / contacts / appointments that must be synchronized to other devices. It should not be editing these, just displaying them when they're relavent.
This shouldn't really be that expensive. Bluetooth headsets are down to the $25 range, so obviously the bluetooth hardware isn't that expensive. And digital watches are way under $20 so that can't be a huge issue. This kind of a feature set is the only reason I'd ever be tempted to buy a watch again, so the sooner the watch industry gets on this, the sooner it makes a potential sale.
I would also take Gentoo over RedHat/Fedora any day of the week. ;) And I understand that some people don't like the way Debian does things. What it comes down to is that *for me* Debian is the right compromise between number of packages, frequency of updates (I use "unstable" on my desktop and "stable" on production servers) and freedom to set my machine up the way I want. Yes, Gentoo gives you slightly more package selection and flexibility in setting up some things. And for some people "the Debian way" is a deal breaker.
It seems like emerge has the ability to easily mix source and binary packages. What I think would be really nice to see is a repository of binary packages for most of the core of the operating system, with the ability to easily force some packages to be compiled from source. For example it would be neat to install one's entire system from binaries; kernel, X11, gnome, mplayer, firefox, etc, then go back and set an override somewhere so that mplayer gets rebuilt from source to take advantage of all the nifty features of your processor. I'm not even saying that *all* packages need to be available as binaries. Heck, you could do a Debian style popularity contest and just keep the most popular packages around as binaries. I've always lived by the "Don't optimize all your code, just optimize the 10% that's being run 90% of the time." In my case I'd like to recompile a couple packages to select a certain non-default feature or enable a certain optimization, but it's not worth devoting the resources (disk space, CPU time, disk I/O) to compile every program I use from source.
While I'm making my wishlist, a program to log which libraries/programs are using the most CPU time, would be cool too. "Oh, I downloaded the binary package for libfoo, and it's pretty CPU intensive and a lot of programs link against it. Maybe I'll recompile it with some optimizations turned on." And, yes I've heard of "top" but I want something that sits in the background and will print out a nice report every so often. Once someone does all that, I'll drop Debian in a heartbeat, but for now it's the OS that annoys me the least.
The new installer looks very slick and I'd love to give this a try. However, I really would prefer to use binary packages. I know that I could get a minor speed improvements compiling by hand, but I have fast enough hardware these days that I'm willing to trade some speed for not having to wait around while Gnome compiles. So, where are these binary packages you speak of? Gentoo.org doesn't seem to list any central repository of pre-compiled packages, and a couple searches on Google have only told me that:
1) Gentoo supports binary packages -and-
2) I should just compile from source because it's better
So, is there an unofficial place to download binary packages that is something like a recognized authority? Where do I get these fabled binary packages? I'm not trolling, I really would like to know.
I know exactly what you mean! Using a hammer just feels like ... cheating.
While I agree that this thing is overpriced, lots of people have Internet access in their kitchens. It's called "wireless" networking.
Cold...Canada looks cold.
Yeah, but it would also be stupid easy to just "lose" that flag while doing a recording. If they don't want the audio to sound like shit, this "flag" has to be outside the human hearing range. If it's outside our hearing range, then just use an analog filter made out of cheap radioshack parts to hack off the flag. Then record it into a device that you *do* have end to end control of, like a tape recorder or any computer made before the great DRM law of 2007. Keep these current generation computers around! They might be useful once DRM is built into the hardware at a firmware level (and required by law...)
Supposedly, eInk screens only use power when changing the image they display. If that's true the batteries really will last for however long it takes you to read 7,500 pages. That is as long as you don't take so long to read 7,500 pages that the self discharge rate of whatever type of batteries it uses becomes a factor...
So, the next question is: What do they consider a good mobile phone?
Actually I think I did see that one...but it was one ESPN8, the "Ocho."
actually booting off of even a "slow" flash memory device, like an older usb drive will be quite quick. Much faster than booting off of a CD-ROM and quite close to the speed of booting off of a hard drive. During a normal boot process you're loading a lot of smallish programs/files, and this plays to the advantages of flash media: no seek times. CD-ROMs have seek times in the tens of milliseconds (maybe even 100 ms for an older unit). Harddrives less than 10 ms these days. Flash media on the other hand is truly random access in the same way that DRAM is, in that there isn't any kind of "seeking" done.