And there's one thing that really bothers me that makes me believe this person didn't read the article.
1) The system is brand new and he said he used the drivers included on the CD since of course Windows wouldn't have drivers for a sound card that didn't exist until later. Isn't having a relatively uniform hardware support system wonderful?
And....uh...I think WinModems were first developed by Rockwell. It sure isn't Intel since they can't develop any decent comm hardware.
If a manufacturer was to supply a driver, how many ways are there to install a driver into Linux versus Windows?
With windows, you always have these inf/dll/vxd bundles which are installed through two faces of the same mechanism "Add/Remove Hardware" and the plug and play autodetect.
With Linux, you have to first choose your sound daemon. http://www.daemons.kiev.ua/multimedia/ Pick one. Which one's better? Beats me. Now compile it into your kernal. Or something like that. (Keep in mind that I ended up giving up on sound when I last tried this cuz I didn't want to figure out what connected where software wise).
Point is, Linux has no unified hardware support architecture. You can't package a driver easily any better than you can package an application. Not only can you not just say "add this driver please" and have the OS handle it, you can't even decide how this stuff's going to arrive. (apt? rpm? tar.gz with only headers and binaries? a library with no api? a text file with pinouts and a few waveforms?)
Not just sound, you can see this problem in the wide variety of X11 implementations too. In Windows, it's add a display driver. In Mac OS Classic, it's add an extension and/or a rom on the video card. In Mac OS X, it's a rom and/or a kext. In Linux? Uh....now what?
Imagine if you can just dump a bunch of say ".drv" packages (which could just be folders with a plist and some custom binary module) into a folder called drivers and have the system check each of them to see if there's a device to match them or just index them for later. Then you can run whatever interface you want on top. There's no reason for each user to learn to tweak XF86Config4 or recompile their kernal for one sound card versus the other. Firewire would just work if there's a driver package detected for your card. RAID cards would only require that you run the configure tool after you boot the first time. etc.... You could switch your interface and hardware without worries. You'll always have a GUI if you want since there's a default piece-of-crap-card driver for video if you haven't downloaded your card's package yet. (yeah, I gave up on XFree86 on a laptop twice already)
In short, there's no unified hardware support architecture and there needs to be one.
The development cycle for USB devices was much too long for the iMac
Wrong. If you've developed USB devices before, you know that all the microcontroller chipsets for making the devices are pretty simple. They come with a HID support already, which takes care of all your input devices. The other mode is bulk transfer which deals with everything else like a serial port except faster. The amount of development time is basically having a product ready, and slapping on a chip with some IDs and routing your driver (if not HID) to use yet another virtual serial port. Development like that can be done in less than a month. (For more info check out Cypress Semiconductor and www.microchip.com)
The only devices availiable before the iMac came out were basically a few mice and keyboards. MS had them, maybe one japanese OEM had them. There were a lot of devices which were announced however but that could mean anything from "oh, it's almost done" to "we're thinking of doing..." to "er, yeah, me too."
Pioneer DVDR drives work fine. The official Superdrive works great at work. The cheap OEM on-sale A04 works great at home. Both of them work with iDVD if you put it in the internal drive bay.
Very true. Apple sure didn't supply silicon to make devices, however, before the iMac, no vendor would have made USB devices since Win95 and Win98 didn't have complete USB implementations back then. And besides, the same developer argument regarding porting from Windows to Linux/Mac applies here. There was no reason for them to make USB devices back which only worked on some machine rather than do it over RS232 or Parallel and make it work with ALL the PCs availiable.
The only incentive for them to make USB devices when the iMac came out was since Mac users have typically tolerated a markup on addons, they were able to charge much much more per unit made AND they were assured they'd be first to market if they got it done fast.
Eventually, if the device wasn't compliant with both the iMac and Win98SE, the device wasn't going to sell.
Wrong..... The Quicktime on Linux project you're referring to is not Apple's Quicktime but rather somebody else's implementation of Quicktime. For the most part, it'd be incomplete in terms of interfaces to hardware devices for audio and video. It's also not compatible with most of the plugin modules and add on features of the real Quicktime.
It's like saying that any add on modules to Windows is binary and API compatible with mplayer. No.
Ah, but note that the parent I was responding to was talking about having it as a CD Changer.
I haven't seen every CD changer but most of the ones I see include the ability to transfer the display data to the head unit, which is what I was commenting on.
Okay, I concede that it's more than just an Aux and charger. It's an aux, charger, and a remote. But adding the remote is almost negligable for me since I could make my own.
If I wanted just to have the control integrated and not all the display info, I'd have just scoped out the lines from the car's remote and then hack it together with an ipod remote for about 3 hours worth of time and $40 (if you ordered a new remote).
And if I was really bored, I got craploads of old microcontrollers. So usng this page: http://www.maushammer.com/systems/ipod-remote/ipod -remote.html I could do without the $40 and have it done "my way" for just time. (Assuming we all had aux and charger, which can be hacked together as well.)
Basically, yes, from the info you've given me, you're probably more likely to be able to do it and all the others I've seen ask me.
You'll have to get the datasheets for the 12" and get the pinouts and/or summary and figure out, for starters, if your panel uses LVDS (look for something like Red Bit 1 Pos followed by a Red Bit 1 Neg in the pinouts).
More information that you might need to check up on include the number of pixel clocks for both the board and the panel. The number of bits per color (if your board supports 8 and the panel supports just 6, you'll just get less colors).
For the most part, single board and embedded computers will have support for more than enough pixel clocks and bits. You'll have to just get firmware to specify what the board to should output.
LVDS stands for Low Voltage Differential Signaling. This means several things to anybody on this article/topic.
First off, LVDS typically does not imply any protocol at all, but rather a category of devices that use two wires (that's where "differential" comes from) per signal (which can be like 8 signals per color as an example) and uses low voltages for those two wires (typically 3.3v for the signal lines, but 3.3, 5, or 12 for the panel main power).
The other common category in LCDs are TTL, which is Transistor Level Logic, and implies that it probably uses lines that are 5 volts for signaling or maybe a wider range.
TTL and LVDS have been used widely to refer to a lot of other devices and their electrical specifications. LVDS SCSI as an example. TTL vs CMOS discrete logic chips. That's it.
Now, knowing that, LVDS in the LCD panel world has come to typically refer to a chip from TI (the number eludes me at the moment). This means that most LCDs marked LVDS, given that you manage to power them up, can actually substitute for each other when connecting to a controller. However, note that for each resolution, the signaling is different. This means that if you have a controller designed for a 17 inch panel running at 1280X1024, you can put on a 18 inch panel at 1280X1024 but you cannot put on a 15 inch at 1024x768 nor a 22 at 1600X1200.
Most current laptops are LVDS.
The ones referred to in the dutch translation with the VESA feature thing is not. It's either custom or TTL. (I haven't checked)
Finally, all this applies for TFTs. DualScans and monochromes are a different story.
The Dennison iceLink thingie doesn't do much more than make an AUX and possibly a charger for your iPod.
I have a Sony head unit which uses Unilink, meaning that the CD changer can report back the title of the track as well, not just control next/prev track.
Since the iPods don't have a easy way to report song and track information, something like this wouldn't happen anytime soon without a change to the iPod firmware (to add networking, probably through the IPoverFW standard) and a firewire network adapter for communicating to the iPod's OS and your head unit.
In the meantime, you might as well settle for an old P5 or G3 sitting in your trunk and using GnUnilink, an open source hardware Unilink adapter.
Keeping in mind that your choice of hardware is ultimately yours, I personally would like to express my opinion that buying a Dell is a bad idea. I used to do tech support for UC Berkeley. And sure, while most of the problems we see are Windows related, a significant number of Dell laptops come by with hardware problems. Badly designed plastic shells for the latest wide screen laptops. Ethernet ports that somehow lose the ability to respond properly to the system. Battery recalls. Power adapter recalls. Media bay latchs that no longer latch.
If you're going to buy a laptop that you want to have stay in one piece, don't buy a Dell.
Heck, if you want to support Linux indirectly, order a Thinkpad from IBM seeing as they have done a lot for the Linux community. See the extra cost as an investment in your hardware lifespan and a company that helps improve Linux.
I can't recall a laptop brand that has demonstrated worse reliability than a Dell.
It looks pretty good from the only screenshot (http://freshrpms.net/apt/synaptic/synaptic.png) I've been able to find. However, the picture looks like something screwed over some UI guidelines with the strange tab/button thing.....though it's probably GTK that's doing it.
It'd be really nice if in addition to all those descriptions, they also listed the command line calls to do the install. That'd actually be really useful......
1) I've been using computers since the IBM XT was considered current. I taught myself batch scripting back then (1986-ish). Now I script primarily in perl and tcsh. And while I'm great at using the command line (a term is the first thing I open when I boot up my machine; heck, I even open most apps from the App folder using the command line), I still don't get apt.
Why?
Sure, your example of "apt-get install mozilla" looks easy. In fact, it's about all I've been able to install with apt. (Oh yeah, pretend my env is still Debian 3) Where do I learn about all the others? How come dselect's quit command seems to not quit immediately? How do I forcequit dselect when I accidentally selected some packages I didn't want? (current method: alternate tty and kill) Where do I get a package list? Is it supposed to be "apt-get install libtar" or "apt-get install libtar-1.2.11" or "apt-get install libtar-1.2" or something else?
It's easy once you know what's there for you to install, but for those who are new to apt, the amount of knowledge required to know what to put after "apt-get install" is akin to me learning.NET, neither of which I have time for.
2) So, uh, what GUI frontend to apt would you recommend? Seriously. I've got a blank x86 box downstairs and I'll give Linux another shot if I can find a fast/minimalist/easy-to-manage distro.
No it isn't. Not all distros include KPackage as part of their default install nor optional package. In fact, I think I've never heard of KPackage before you mentioned it in this post.
People are used to using a browser and clicking things to download them. Most DMGs (internet enabled DMGs) now mount by themselves and extract the program and leave it in place automatically. As for copying to the Apps folder, it's assumed that the Apps folder is a place the user will learn about fairly quickly seeing that double-clicking the hard drive will reveal it and the hard drive would logically be the best place to start when you first start out on a new computer because your hard drive is where everything is stored.
Besides, who said you needed to copy it to the apps folder? I actually have some apps that I leave on my desktop. No, not shortcuts, but the entire app since it's more convienent for me.
Er......WTF?
How did HP and MS get on the same side and Intel/Dell get on the same side as IBM?
If you're going to pair stuff, at least get the basics right. HP/Linux. IBM/Linux. Intel/Dell. HP and IBM can mix despite having their own CPU archs since HP's Itanium and Alpha tech is pretty much history. MS and HP don't mix since HP's trying to become the next Apple....or Gateway, depending on how they play the digital lifestyle and consumer electronics game. Dell uses exclusively Intel stock tech now and has always stuck with MS products. Just because Dell now has a few linux machine doesn't mean anything, it's just to get around businesses paying MS twice for licences.
I used to also admin a small startup company.
The reason we used our Athlon XPs and P4s with NT4 is that it worked better with our dev tools. We developed hardware, so.NET vs MFC debates don't apply here.
Anyhow, not only did our high end machines boot up insanely fast, they worked exactly like how we expected them to since we ran all the previous machines on NT4 as well and knew what we needed already. The old ones got relegated to doing net-wide backups and for the engineers who thought their machines were fast enough.
Sure, XP is probably compiled with optimizations that let it run faster on a P4, but when XP is also more than twice as big for just the OS install, packed with CPU wasting features (it seems they consider this a service), and running on hardware it was designed for, it's going to be much slower than NT4 on the same hardware. I mean, think about it. NT4 is running on a machine several magnitudes faster than what it was designed for. Tossing in a few new processor optimizations can't account for that.
Oh, and in our evals, XP tended to be more unstable and irrecoverable than NT4 when it decides to crap out.
What are you smoking?
Apparently something not as strong as you are.
The Win95/98 family (ME is like XP, something you must avoid) compared to Win2k, while the kernal is older and not as advanced, take up less resources and is easier to secure and manage compared to Win2k.
Here's why:
1) Win98/95 has no services. This means that when there is spyware/trojans, it's gotta run through the normal startup techniques which are all accessable through msconfig and are deleteable by the user. Ever had DarkIRC go ape on your 2k box? Didn't it annoy you that the first 100 times you cleaned it, the service entries were still there because you didn't know how Firedaemon worked? (BTW, I was onsite tech support at UC Berkeley: www.rescomp.berkeley.edu)
2) No net services, or at least as little as possible also means that there's comparatively little open ports to secure which is just quicker. And if somebody did root your box by remote, only old-school hackers would be able to do much with it compared with all the script kiddies who are now familar with 2k/XP.
3) There's just less stuff running overall. Much faster. It was designed for older hardware which had less to work with. Afterall, a 500 mhz machine was top of the line back then. It's the same kind of magnitude as running 2k on a 3ghz dual with a fat RAID setup.
4) Win2k's old hardware support is spotty. In fact, with my old Athlon 550, Win2k decides that it doesn't like hard drives which can't support at least UDMA33 and nukes the data beyond repair on anything older. Not a good first impression. It also crashed reliably over 20 times in attempted installs on hardware which worked perfectly fine under Win98 and other OSs. This is due to the fact that 2k was a major release after NT4, and you know what happens when MS tries to dump a bunch of new features into a new release....... it doesn't work right.
5) Drivers which worked for 95 will usually work in 98. Since we are talking about OLD hardware, it's obvious that there's going to be hardware which is old enough not to have NT compatible drivers.
Overall, using Win98SE is your best bet for old hardware.
There's MUCH more (including adding and removing RAM without rebooting--
Hold on.... Name me 5 mobos which have the registers and filters in place to safely add in a DIMM without having the power line and signal line noise fux0r all the data that's going along the other DIMMS.
And if you can't, then name one chipset which has, in some way, feedback to tell the OS "oh hey, there's some more memory that wasn't here before."
By the way, OpenGL is from SGI. It's not something that sprung out of the Open Source Movement.
Likewise, on the topic of browsers, while I don't recall the history of Mosaic, you're probably referring to Mozilla which was not open source until after Netscape was beaten down by IE.
And Google could have been built on any platform, it's just that using Linux and some already made OSS packages made their development faster.
That said, the results of the Open Source movement are "original" in terms of projects and implementation. However, a majority of projects are half-assed attempts at duplicating
the functionality of other projects just to say there's a free version.
What I think the person you were replying to is referring that there's not too much in terms of original ideas or mind-blowingly good pieces of software. Mplayer is great, I use that all the time. Likewise VideoLan when Mplayer breaks. It's great that the kernal is flexible enough to be used on massive multiproc servers as well as a tiny router.
But there's a lot that isn't so great. In terms of drivers, I still don't understand how to add drivers. No, I don't want to compile support for my hardware into my kernal everytime I add something. How about installs/uninstalls? There's apt, yum, rpms, etc..... What do I use? How do I use them?.......together? Why isn't there better debugging info for XFree86 when my video card is supported but still causing a kernal panic? Why does everybody try to make a half-assed version of the OSX GUI and actually try to tell me theirs looks just as good when it clearly doesn't adhere to the UI guidelines they're copying? (I run Jaguar, btw) For the most part, all I seem to see are projects for window managers who want to be like another OS, taskbars that want to be like Explorer or Dock.app, Office suites which have a key goal to look like and work like MS Office (not a good thing), talk about "it's just like Windows Update" (not a good thing either), and passing it all off as good because it's free.
There's a reason why it's been hard for me to convince friends, who are coders even, to switch to Linux when they're frustrated with Windows. It's because while there's so much choice, they don't care. They just want stuff to work. And so they all go out and buy Powerbooks and run OSX. I'm certainly happy that they're using OSX (I also am currently employed by a subsidiary of Apple, btw) but it pains me to see that the efforts of people who seem to have a sincere want to bring a better platform to people who can't afford Macs go to waste simply because there's no direction, no killer-app, no ease-of-use, and no coherence to the Linux-on-desktop movement.
In the end, for all my older hardware, I want a Linux distro which only has drivers for standard BIOS boot devices built in, loads all the drivers off a plugin system, has a Xserver which is capable of autodetecting the chipset and monitor attributes (even on laptops) or a unified database of system data which gives you all the known correct settings for each videocard/platform/laptop/system and has a fail-safe default to a simple 640X480 16 color mode so it'll always come up CORRECTLY in a GUI if you do startx or whatever (I have a laptop marked "supported" by name, in multiple linux distros, manages to either come up with a bad palette or a kernal panic), and finally a window manager that is fast, efficient, and requires only one mouse button to access all the features (not because my mouse, which has more than 1 button, but because it's easier.) If somebody can reach that goal, the next step is to make a packaging system where you can install/remove by dragging a folder or even a single app icon into a trash or any location you want it to run from. In short, I'm describing something that vaguely resembles OSX without the flashy look, just a purely simple UI that lets you manage the machine without being able to memorize the entire HOW-TO database.
Freakin' a.....
You guys gotta realize that ATA/66 is enough for most of your rigs now cuz your crappy 7200 rpm drives can't flood that bus anyways. Hell, my 10k rpm server drives can't do it. Getting ATA/133 over ATA/66 is pretty much useless but it's great marketing.
It's kinda like this. You got the transmission system of a Subaru WRX sti rally car. But you got a 1.5 HP lawnmower engine under your hood.
Cubic6 is pretty much right. The latency will kill you more. Mod him up. Heck, while your mentioning bandwidth; ATA/100 bandwidth is pretty much useless anyhow since the hardware of the drive can't dump more than 30 Megabytes/sec. Yeah, sure the cache and track buffers empty nice and fast but your OS should be supplying a fat disk cache already, so essentially those of you saying "my SATA subsystem will own your ATA/133 setup" isn't really faster than ATA/66....which is about to be owned by my quad 10k rpm wide SCSI RAID system.....With each pair on their own independent 40 megabyte/sec channel.
> "how many other servers have a FireWire port on the *front*?"
Just to let you know..... they're useful for quickly booting off a external drive. How many other servers were designed with the consideration to boot off something else in an emergency?
Slashdot Mirror
My powerbook is named Decker, after the class of people in Shadowrun who run around and plug into networks and use them.
My windows XP box..... I'm thinking of renaming it "bitch."
And there's one thing that really bothers me that makes me believe this person didn't read the article.
1) The system is brand new and he said he used the drivers included on the CD since of course Windows wouldn't have drivers for a sound card that didn't exist until later. Isn't having a relatively uniform hardware support system wonderful?
And....uh...I think WinModems were first developed by Rockwell. It sure isn't Intel since they can't develop any decent comm hardware.
Whoa there; take a different angle.
If a manufacturer was to supply a driver, how many ways are there to install a driver into Linux versus Windows?
With windows, you always have these inf/dll/vxd bundles which are installed through two faces of the same mechanism "Add/Remove Hardware" and the plug and play autodetect.
With Linux, you have to first choose your sound daemon. http://www.daemons.kiev.ua/multimedia/ Pick one.
Which one's better? Beats me. Now compile it into your kernal. Or something like that. (Keep in mind that I ended up giving up on sound when I last tried this cuz I didn't want to figure out what connected where software wise).
Point is, Linux has no unified hardware support architecture. You can't package a driver easily any better than you can package an application. Not only can you not just say "add this driver please" and have the OS handle it, you can't even decide how this stuff's going to arrive. (apt? rpm? tar.gz with only headers and binaries? a library with no api? a text file with pinouts and a few waveforms?)
Not just sound, you can see this problem in the wide variety of X11 implementations too. In Windows, it's add a display driver. In Mac OS Classic, it's add an extension and/or a rom on the video card. In Mac OS X, it's a rom and/or a kext. In Linux? Uh....now what?
Imagine if you can just dump a bunch of say ".drv" packages (which could just be folders with a plist and some custom binary module) into a folder called drivers and have the system check each of them to see if there's a device to match them or just index them for later. Then you can run whatever interface you want on top. There's no reason for each user to learn to tweak XF86Config4 or recompile their kernal for one sound card versus the other. Firewire would just work if there's a driver package detected for your card. RAID cards would only require that you run the configure tool after you boot the first time. etc.... You could switch your interface and hardware without worries. You'll always have a GUI if you want since there's a default piece-of-crap-card driver for video if you haven't downloaded your card's package yet. (yeah, I gave up on XFree86 on a laptop twice already)
In short, there's no unified hardware support architecture and there needs to be one.
The development cycle for USB devices was much too long for the iMac
Wrong. If you've developed USB devices before, you know that all the microcontroller chipsets for making the devices are pretty simple. They come with a HID support already, which takes care of all your input devices. The other mode is bulk transfer which deals with everything else like a serial port except faster. The amount of development time is basically having a product ready, and slapping on a chip with some IDs and routing your driver (if not HID) to use yet another virtual serial port. Development like that can be done in less than a month. (For more info check out Cypress Semiconductor and www.microchip.com)
The only devices availiable before the iMac came out were basically a few mice and keyboards. MS had them, maybe one japanese OEM had them. There were a lot of devices which were announced however but that could mean anything from "oh, it's almost done" to "we're thinking of doing..." to "er, yeah, me too."
Pioneer DVDR drives work fine. The official Superdrive works great at work. The cheap OEM on-sale A04 works great at home. Both of them work with iDVD if you put it in the internal drive bay.
Very true. Apple sure didn't supply silicon to make devices, however, before the iMac, no vendor would have made USB devices since Win95 and Win98 didn't have complete USB implementations back then. And besides, the same developer argument regarding porting from Windows to Linux/Mac applies here. There was no reason for them to make USB devices back which only worked on some machine rather than do it over RS232 or Parallel and make it work with ALL the PCs availiable.
The only incentive for them to make USB devices when the iMac came out was since Mac users have typically tolerated a markup on addons, they were able to charge much much more per unit made AND they were assured they'd be first to market if they got it done fast.
Eventually, if the device wasn't compliant with both the iMac and Win98SE, the device wasn't going to sell.
Wrong..... The Quicktime on Linux project you're referring to is not Apple's Quicktime but rather somebody else's implementation of Quicktime. For the most part, it'd be incomplete in terms of interfaces to hardware devices for audio and video. It's also not compatible with most of the plugin modules and add on features of the real Quicktime.
It's like saying that any add on modules to Windows is binary and API compatible with mplayer. No.
Ah, but note that the parent I was responding to was talking about having it as a CD Changer.
d -remote.html
I haven't seen every CD changer but most of the ones I see include the ability to transfer the display data to the head unit, which is what I was commenting on.
Okay, I concede that it's more than just an Aux and charger. It's an aux, charger, and a remote. But adding the remote is almost negligable for me since I could make my own.
If I wanted just to have the control integrated and not all the display info, I'd have just scoped out the lines from the car's remote and then hack it together with an ipod remote for about 3 hours worth of time and $40 (if you ordered a new remote).
And if I was really bored, I got craploads of old microcontrollers. So usng this page: http://www.maushammer.com/systems/ipod-remote/ipo
I could do without the $40 and have it done "my way" for just time. (Assuming we all had aux and charger, which can be hacked together as well.)
Basically, yes, from the info you've given me, you're probably more likely to be able to do it and all the others I've seen ask me. You'll have to get the datasheets for the 12" and get the pinouts and/or summary and figure out, for starters, if your panel uses LVDS (look for something like Red Bit 1 Pos followed by a Red Bit 1 Neg in the pinouts). More information that you might need to check up on include the number of pixel clocks for both the board and the panel. The number of bits per color (if your board supports 8 and the panel supports just 6, you'll just get less colors). For the most part, single board and embedded computers will have support for more than enough pixel clocks and bits. You'll have to just get firmware to specify what the board to should output.
LVDS stands for Low Voltage Differential Signaling.
This means several things to anybody on this article/topic.
First off, LVDS typically does not imply any protocol at all, but rather a category of devices that use two wires (that's where "differential" comes from) per signal (which can be like 8 signals per color as an example) and uses low voltages for those two wires (typically 3.3v for the signal lines, but 3.3, 5, or 12 for the panel main power).
The other common category in LCDs are TTL, which is Transistor Level Logic, and implies that it probably uses lines that are 5 volts for signaling or maybe a wider range.
TTL and LVDS have been used widely to refer to a lot of other devices and their electrical specifications. LVDS SCSI as an example. TTL vs CMOS discrete logic chips. That's it.
Now, knowing that, LVDS in the LCD panel world has come to typically refer to a chip from TI (the number eludes me at the moment). This means that most LCDs marked LVDS, given that you manage to power them up, can actually substitute for each other when connecting to a controller. However, note that for each resolution, the signaling is different. This means that if you have a controller designed for a 17 inch panel running at 1280X1024, you can put on a 18 inch panel at 1280X1024 but you cannot put on a 15 inch at 1024x768 nor a 22 at 1600X1200.
Most current laptops are LVDS.
The ones referred to in the dutch translation with the VESA feature thing is not. It's either custom or TTL. (I haven't checked)
Finally, all this applies for TFTs. DualScans and monochromes are a different story.
Not like that.
The Dennison iceLink thingie doesn't do much more than make an AUX and possibly a charger for your iPod.
I have a Sony head unit which uses Unilink, meaning that the CD changer can report back the title of the track as well, not just control next/prev track.
Since the iPods don't have a easy way to report song and track information, something like this wouldn't happen anytime soon without a change to the iPod firmware (to add networking, probably through the IPoverFW standard) and a firewire network adapter for communicating to the iPod's OS and your head unit.
In the meantime, you might as well settle for an old P5 or G3 sitting in your trunk and using GnUnilink, an open source hardware Unilink adapter.
Oh man.....
Keeping in mind that your choice of hardware is ultimately yours, I personally would like to express my opinion that buying a Dell is a bad idea. I used to do tech support for UC Berkeley. And sure, while most of the problems we see are Windows related, a significant number of Dell laptops come by with hardware problems. Badly designed plastic shells for the latest wide screen laptops. Ethernet ports that somehow lose the ability to respond properly to the system. Battery recalls. Power adapter recalls. Media bay latchs that no longer latch.
If you're going to buy a laptop that you want to have stay in one piece, don't buy a Dell.
Heck, if you want to support Linux indirectly, order a Thinkpad from IBM seeing as they have done a lot for the Linux community. See the extra cost as an investment in your hardware lifespan and a company that helps improve Linux.
I can't recall a laptop brand that has demonstrated worse reliability than a Dell.
It looks pretty good from the only screenshot (http://freshrpms.net/apt/synaptic/synaptic.png) I've been able to find. However, the picture looks like something screwed over some UI guidelines with the strange tab/button thing.....though it's probably GTK that's doing it.
It'd be really nice if in addition to all those descriptions, they also listed the command line calls to do the install. That'd actually be really useful......
Anyhow, I'll give it a shot this weekend. Thanks
(BTW: I primarily use a Powerbook.)
.NET, neither of which I have time for.
1) I've been using computers since the IBM XT was considered current. I taught myself batch scripting back then (1986-ish). Now I script primarily in perl and tcsh. And while I'm great at using the command line (a term is the first thing I open when I boot up my machine; heck, I even open most apps from the App folder using the command line), I still don't get apt.
Why?
Sure, your example of "apt-get install mozilla" looks easy. In fact, it's about all I've been able to install with apt. (Oh yeah, pretend my env is still Debian 3) Where do I learn about all the others? How come dselect's quit command seems to not quit immediately? How do I forcequit dselect when I accidentally selected some packages I didn't want? (current method: alternate tty and kill) Where do I get a package list? Is it supposed to be "apt-get install libtar" or "apt-get install libtar-1.2.11" or "apt-get install libtar-1.2" or something else?
It's easy once you know what's there for you to install, but for those who are new to apt, the amount of knowledge required to know what to put after "apt-get install" is akin to me learning
2) So, uh, what GUI frontend to apt would you recommend? Seriously. I've got a blank x86 box downstairs and I'll give Linux another shot if I can find a fast/minimalist/easy-to-manage distro.
No it isn't. Not all distros include KPackage as part of their default install nor optional package. In fact, I think I've never heard of KPackage before you mentioned it in this post.
People are used to using a browser and clicking things to download them. Most DMGs (internet enabled DMGs) now mount by themselves and extract the program and leave it in place automatically. As for copying to the Apps folder, it's assumed that the Apps folder is a place the user will learn about fairly quickly seeing that double-clicking the hard drive will reveal it and the hard drive would logically be the best place to start when you first start out on a new computer because your hard drive is where everything is stored.
Besides, who said you needed to copy it to the apps folder? I actually have some apps that I leave on my desktop. No, not shortcuts, but the entire app since it's more convienent for me.
Er......WTF?
How did HP and MS get on the same side and Intel/Dell get on the same side as IBM?
If you're going to pair stuff, at least get the basics right. HP/Linux. IBM/Linux. Intel/Dell. HP and IBM can mix despite having their own CPU archs since HP's Itanium and Alpha tech is pretty much history. MS and HP don't mix since HP's trying to become the next Apple....or Gateway, depending on how they play the digital lifestyle and consumer electronics game. Dell uses exclusively Intel stock tech now and has always stuck with MS products. Just because Dell now has a few linux machine doesn't mean anything, it's just to get around businesses paying MS twice for licences.
I used to also admin a small startup company. .NET vs MFC debates don't apply here.
The reason we used our Athlon XPs and P4s with NT4 is that it worked better with our dev tools. We developed hardware, so
Anyhow, not only did our high end machines boot up insanely fast, they worked exactly like how we expected them to since we ran all the previous machines on NT4 as well and knew what we needed already. The old ones got relegated to doing net-wide backups and for the engineers who thought their machines were fast enough.
Sure, XP is probably compiled with optimizations that let it run faster on a P4, but when XP is also more than twice as big for just the OS install, packed with CPU wasting features (it seems they consider this a service), and running on hardware it was designed for, it's going to be much slower than NT4 on the same hardware. I mean, think about it. NT4 is running on a machine several magnitudes faster than what it was designed for. Tossing in a few new processor optimizations can't account for that.
Oh, and in our evals, XP tended to be more unstable and irrecoverable than NT4 when it decides to crap out.
What are you smoking?
Apparently something not as strong as you are.
The Win95/98 family (ME is like XP, something you must avoid) compared to Win2k, while the kernal is older and not as advanced, take up less resources and is easier to secure and manage compared to Win2k.
Here's why:
1) Win98/95 has no services. This means that when there is spyware/trojans, it's gotta run through the normal startup techniques which are all accessable through msconfig and are deleteable by the user. Ever had DarkIRC go ape on your 2k box? Didn't it annoy you that the first 100 times you cleaned it, the service entries were still there because you didn't know how Firedaemon worked? (BTW, I was onsite tech support at UC Berkeley: www.rescomp.berkeley.edu)
2) No net services, or at least as little as possible also means that there's comparatively little open ports to secure which is just quicker. And if somebody did root your box by remote, only old-school hackers would be able to do much with it compared with all the script kiddies who are now familar with 2k/XP.
3) There's just less stuff running overall. Much faster. It was designed for older hardware which had less to work with. Afterall, a 500 mhz machine was top of the line back then. It's the same kind of magnitude as running 2k on a 3ghz dual with a fat RAID setup.
4) Win2k's old hardware support is spotty. In fact, with my old Athlon 550, Win2k decides that it doesn't like hard drives which can't support at least UDMA33 and nukes the data beyond repair on anything older. Not a good first impression. It also crashed reliably over 20 times in attempted installs on hardware which worked perfectly fine under Win98 and other OSs. This is due to the fact that 2k was a major release after NT4, and you know what happens when MS tries to dump a bunch of new features into a new release....... it doesn't work right.
5) Drivers which worked for 95 will usually work in 98. Since we are talking about OLD hardware, it's obvious that there's going to be hardware which is old enough not to have NT compatible drivers.
Overall, using Win98SE is your best bet for old hardware.
There's MUCH more (including adding and removing RAM without rebooting-- Hold on.... Name me 5 mobos which have the registers and filters in place to safely add in a DIMM without having the power line and signal line noise fux0r all the data that's going along the other DIMMS. And if you can't, then name one chipset which has, in some way, feedback to tell the OS "oh hey, there's some more memory that wasn't here before."
By the way, OpenGL is from SGI. It's not something that sprung out of the Open Source Movement. Likewise, on the topic of browsers, while I don't recall the history of Mosaic, you're probably referring to Mozilla which was not open source until after Netscape was beaten down by IE. And Google could have been built on any platform, it's just that using Linux and some already made OSS packages made their development faster. That said, the results of the Open Source movement are "original" in terms of projects and implementation. However, a majority of projects are half-assed attempts at duplicating the functionality of other projects just to say there's a free version. What I think the person you were replying to is referring that there's not too much in terms of original ideas or mind-blowingly good pieces of software. Mplayer is great, I use that all the time. Likewise VideoLan when Mplayer breaks. It's great that the kernal is flexible enough to be used on massive multiproc servers as well as a tiny router. But there's a lot that isn't so great. In terms of drivers, I still don't understand how to add drivers. No, I don't want to compile support for my hardware into my kernal everytime I add something. How about installs/uninstalls? There's apt, yum, rpms, etc..... What do I use? How do I use them?.......together? Why isn't there better debugging info for XFree86 when my video card is supported but still causing a kernal panic? Why does everybody try to make a half-assed version of the OSX GUI and actually try to tell me theirs looks just as good when it clearly doesn't adhere to the UI guidelines they're copying? (I run Jaguar, btw) For the most part, all I seem to see are projects for window managers who want to be like another OS, taskbars that want to be like Explorer or Dock.app, Office suites which have a key goal to look like and work like MS Office (not a good thing), talk about "it's just like Windows Update" (not a good thing either), and passing it all off as good because it's free. There's a reason why it's been hard for me to convince friends, who are coders even, to switch to Linux when they're frustrated with Windows. It's because while there's so much choice, they don't care. They just want stuff to work. And so they all go out and buy Powerbooks and run OSX. I'm certainly happy that they're using OSX (I also am currently employed by a subsidiary of Apple, btw) but it pains me to see that the efforts of people who seem to have a sincere want to bring a better platform to people who can't afford Macs go to waste simply because there's no direction, no killer-app, no ease-of-use, and no coherence to the Linux-on-desktop movement. In the end, for all my older hardware, I want a Linux distro which only has drivers for standard BIOS boot devices built in, loads all the drivers off a plugin system, has a Xserver which is capable of autodetecting the chipset and monitor attributes (even on laptops) or a unified database of system data which gives you all the known correct settings for each videocard/platform/laptop/system and has a fail-safe default to a simple 640X480 16 color mode so it'll always come up CORRECTLY in a GUI if you do startx or whatever (I have a laptop marked "supported" by name, in multiple linux distros, manages to either come up with a bad palette or a kernal panic), and finally a window manager that is fast, efficient, and requires only one mouse button to access all the features (not because my mouse, which has more than 1 button, but because it's easier.) If somebody can reach that goal, the next step is to make a packaging system where you can install/remove by dragging a folder or even a single app icon into a trash or any location you want it to run from. In short, I'm describing something that vaguely resembles OSX without the flashy look, just a purely simple UI that lets you manage the machine without being able to memorize the entire HOW-TO database.
Freakin' a..... You guys gotta realize that ATA/66 is enough for most of your rigs now cuz your crappy 7200 rpm drives can't flood that bus anyways. Hell, my 10k rpm server drives can't do it. Getting ATA/133 over ATA/66 is pretty much useless but it's great marketing. It's kinda like this. You got the transmission system of a Subaru WRX sti rally car. But you got a 1.5 HP lawnmower engine under your hood.
Cubic6 is pretty much right. The latency will kill you more. Mod him up. ....With each pair on their own independent 40 megabyte/sec channel.
Heck, while your mentioning bandwidth; ATA/100 bandwidth is pretty much useless anyhow since the hardware of the drive can't dump more than 30 Megabytes/sec. Yeah, sure the cache and track buffers empty nice and fast but your OS should be supplying a fat disk cache already, so essentially those of you saying "my SATA subsystem will own your ATA/133 setup" isn't really faster than ATA/66....which is about to be owned by my quad 10k rpm wide SCSI RAID system.
> "how many other servers have a FireWire port on the *front*?" Just to let you know..... they're useful for quickly booting off a external drive. How many other servers were designed with the consideration to boot off something else in an emergency?