And I'm referring to the article and the guy who decided to do this "experiment."
(1) This software is designed for you to be able to license content licensable by AP. It assumes that the content you want to license was gotten from AP, because only people who want to license from AP will fill out the form to license text from AP. DUH. (2) Technically, It may be prohibitive for them to write their software to sift through terabytes of text for every license request to ensure that the text you asked to license is (a) in there and (b) not licensed from someone else or public domain. (3) It's fucking easy for anyone to guess that their software would behave this way. (4) It's fucking unnecessary for it to be "smart" enough to second-guess stupid people who try to do this. (5) Anyone knowingly trying to license from AP something not owned by AP is being a jerk. (6) Anyone unknowingly trying to license from AP something not owned by AP is an idiot because they didn't do their homework. (7) If you're a competent journalist, you'll know your rights, research the origin of content you want to quote, and attribute and license it appropriately. That's YOUR responsibility.
We all like a laugh at the expense of the **AA or the various media outlets. We are righteously outraged when they knowingly claim ownership of content that is not theirs. We are righteously outraged when they issue take-down notices for things that are obviously fair use.
But this is just fucking useless and stupid. I'm sorry. I'm pissed about this guy wasting our time with this, and some of you jackasses are swallowing it hook, line, and sinker. What pisses me off is that organizations like the AP are doing all kinds of things that REALLY ARE unfair, and yet some dick decides he's going to start picking on a non-issue. Now WE'RE the jackasses who are harrassing the AP. Two wrongs don't make a right.
I see nothing wrong with this service. I think it's perfectly sensible. The AP is just trying to save themselves and everyone else time and money by developing a convenient automated system for licensing content. Doing it this way saves on all sorts of legal, technical, and manpower costs. If they were to put in the effort to develop a system that would second-guess idiots who try to license content that's not actually under AP's control, it would cost ass-loads of money, and the service would be slow as shit. It wouldn't be worth the expense, so they'd shut it down and return to the old, expensive, manual way of doing things. You can argue about whether or not it's right for them to charge AT ALL for content. But that's a separate issue. Courts will uphold that they DO have a right to license at least SOME of their content, and this automated system is helping them keep volumes up and individual license fees down.
If you really want to catch the AP with their pants down, you'll have to find a situation (and I'm sure there are many) where the AP have explicitly (with a human involved) claimed ownership of content that wasn't theirs. In the case discussed in this article, I don't think it would stand up in any court that their automated system was being malicious, because the fact that they charged for a license doesn't necessarily imply ownership. Nowhere in any of this is any intent to defraud or deceive. The only way for you to be tricked by this system is to trick yourself.
Back when I was young (I graduated highschool in 1991), I recall people who migrated from WordPerfect to Word complaining about the missing "reveal codes" option. I looked into this, and this is what my friends with Ph.D.s at the time told me: Word didn't have "reveal codes" because it didn't have codes.
Let me step back and explain this a little better. Word Perfect used in-line codes to indicate formatting. There was an "italics on" code and there was an "italics off" command. It's not quite like HTML or XML, because it wasn't hierarchical. A document was a linear stream of bytes, and the word processor displayed the formatting by traversing the bytes to figure it out. On the processors of the day (386's), this had some major performance disadvantages, when the program had to scan back thousands of bytes just to figure out what the correct formatting was for what was being displayed on the screen. This was okay for the DOS version (can't see most of the formatting, so don't need to look for it), but it became a major liability for the Windows version. It was also a liability because documents that had been edited and edited tended to crud up with lots of superflous codes that WP simply didn't have the smarts to clean up. The only "advantage" was that you could reveal the codes, and that was only an advantage because people got used to it, and they got used to it because WP became problematic to use if you didn't reveal the codes to clean up problems.
Word did things differently. We all like to complain about Microsoft's behavior, and we like to complain about how crufty their software is. But now and then, their engineers (who are people like anyone else) did manage to do something that had intelligence behind it. Mind you, sometimes something has intelligence simply because someone thought about it and made an engineering decision. I'm not trying to claim that this was necessarily BETTER. Anyhow, Word didn't have reveal codes because it didn't have codes, per se, to reveal. Not in-line anyhow. Word was object-oriented. Word documents contained data structures that themselves indicated formatting and contained text. Paragraphs were objects. Sections were objects. Text within italics was inside an object. In a way, this is neither here nor there compared to reveal codes, but it made a practical difference in that when Word needed to determine the formatting of an object, rather than scanning back to the beginning of the file (which WP didn't always have to do but did sometimes which made it slow), Word worked its way up the object hierarchy, a much more efficient process. This also had advantages in that the object tree could be optimized to contain the formatting that was actually there. In WP, if you un-italicized a sentence that had been italicized, it wouldn't necessarily remove the old codes, instead inserting extra codes so that you got on's followed immediately by off's. Word would just delete the object.
So, to summarize, the reason Word didn't have reveal codes was that there were no in-line codes to reveal. Word's equivalent would have been some way to display the object hierarchy, which wouldn't necessarily have been intuitively useful to users. And of course, it would have been silly to emulate codes just to imitate a "feature" of WP that only existed in the first place because WP didn't automatically manage its codes properly.
We like to attribute scientific foresight to sci-fi. And indeed, sci-fi has been the inspiration for many a scientist or engineer. Star Trek writers probably picked aluminum because of its high strength and low density. But unless there was some kind of prior scientific advancement that they knew about, the choice of aluminum amounted to little more than a convenient plot device. It doesn't seem implausible to me that it would foreshadow the development of SOME kind of metal or other non-SiO2 material that's light, strong, and transparent. But not necessarily aluminum.
My office-mates were discussing this. Apparently, some professor on the radio was complaining that kids would be unable to read older documents written in cursive. And that seemed the best argument he could come up with.
Honestly, I think that's silly. They don't teach gothic script (Sütterlinschrift or Fraktur) in Germany anymore. Those who need it just LEARN it. This is no different from learning to read Chinese, Hebrew or particularly Arabic. They're not THAT hard to learn when you need them.
But maybe that's just me. Do most people have a really hard time learning to interpret a foreign script? It seems to me that it should be particularly easy if it's just a "code" for the same letters and words you use in your native language.
I wonder how well the space agencies are taking into consideration human need for intimate contact. NASA crews at least get some scheduled parts of the day when they get privacy and are disconnected from health monitors. But this is for relatively short missions. Unsatisfied sexual needs can create tension, especially in situations of extreme isolation with no way out. Perhaps only sexually active couples should be sent on these trips?
With respect to actually knowing things and getting work done, it's better to understand computer science concepts in general then to know obscure details of one or a few particular languages. But keep in mind that the people who hire progammers are not themselves programmers. Head hunters and HR departments are generally non-technical and will not understand (or care) if you talk about these concepts. The fact is, unless you're getting a Ph.D., you have to tailor your resume to be consumed by a half-broken computer program with multiple choice check boxes, and that is how you're going to be judged. It doesn't matter if you know C++, Java, Lisp, Ruby, PHP5, Python, Haskell, and assembly for 10 different processors. If the job calls for C#, and you don't claim to know it, you won't even be considered. If you DO claim to know it, it's hit or miss whether or not you'll be tested on it. I've actually gotten less than perfect scores on C and C++ tests due to things that I KNOW are wrong with the test, so you're up against that too. As much as you have to know computer science, you also have to know how to game the system.
Here's what has worked for me:
(1) Know your concepts well. Know how to take any program and implement it in pseudocode. Learn how to do it in structured, object oriented, and functional languages. This will help you actually get the job done and get raises and stuff once you're hired. (2) Find out what languages are hot and read at least one small tutorial on each, and write programs in each of at least, say, 1000 lines. Whatever it takes to get used to the basics. Also, read the wikipedia articles and google for "trick and tips" and "gochas" and other things that people praise or gripe about for those languages. Get your feet wet. This way, you can then claim that you have programmed in each of those languages, and you can answer some of the weirder questions someone might ask. (3) Familiarize yourself with different types programming environments or platforms. Program something on Windows. Program something on a Mac (if you can get your hands on one easily enough). Program something on Linux. Program something on an embedded system or at least a language used on them.
You have to lie judiciously. For instance, my background included assembly for several processors, C, C++, for modern system as well as things like 6502's, along with a fair amount of chip design experience. Could I do embedded systems programming? Duh. But at this one point, I didn't have any ACTUAL embedded systems programming experience. So I was honest and explained that while I hadn't, I clearly had enough related experience that was easy. I didn't get the job. What I should have done was done a bit of reading on the subject to ensure that I know the names of a few embedded processors and bluffed my way through. The problem was that the head hunter wasn't technical enough to understand my explanation, so all he knew to do was check "NO" on the embedded systems experience. He can comment in there about the chip design, but the HR people he would hand this off to wouldn't know what do do with it.
To summarize, to prepare yourself for working in the IT industry, you have to learn programming, but more importantly, you have to learn how to translate what you know into the language of the nontechnical people who do the hiring. This requires a kind of intelligence (subterfuge, to a limited extent) that many technophiles are not very skilled at. If you don't you give away control to the people who understand the art of deception better than they understand technical things. But those are the ones who rule the world. Politicians succeed not by knowing things but by knowing how to SPIN things.
There's an mdadm RAID5 driver, but I don't think dmraid does anything other than RAID0 and RAID1.
For RAID1, there's an optimization that one can make, which is to distribute reads between the two mirrors. For RAID0 and RAID5, there's only one place that any disk block can be read from, so there's no room for optization (or lack thereof).
See my other posts on this article. The Linux kernel has two separate software RAID drivers. "dmraid" interfaces with BIOS-based software RAID, while "mdraid" is Linux's native software RAID. There are two separate drivers. Unless you're trying to maintain compatibility with Windows using BIOS RAID, then dmraid is the wrong one to use. Use mdraid.
You are misinformed. dmraid (fakeraid) and mdraid (Linux native software RAID) _should_ use the same driver. The only real difference is where the metadata is stored (dmraid asks the BIOS, mdraid handles it in a more open way). Once the metadata is loaded, they're both just the OS accessing regular disks. What's the difference?
But in fact, there are two separate drivers. I discovered this by accidentally choosing the wrong one. When I discovered the difference (random reads are faster on mdraid but not dmraid), I asked on LKML and got no response. So I went digging through the kernel source myself and discovered the two separate drivers. The mdraid does a simple round-robbin. Whenever the next request is not contiguous from the last one (block number plus one), it switches to the other disk. No more intelligence than that. The dmraid, on the other hand, directs all reads to just the primary disk.
Heh. Linux software RAID doesn't do jack. I've looked at the source code. The mdadm RAID1 driver just alternates drives for reads whenever the requests are not contiguous. That is all. Nothing more. There's no intelligence in there. No keeping track of head positions, no attempts to discover or infer physical drive geometry. Nothing. Just a simple round-robbin. It just so happens that for MOST things that involve random access, the effective throughput is nearly doubled. More intelligence wouldn't actually buy you much in the general case, so why bother?
Also, the dmraid (fakeraid) RAID1 driver only does reads from one disk. I made the mistake of using dmraid instead of mdraid, only to discover through performance tests and iostat that there are basically two software RAID drivers that CLAIM to do identical things but in fact do not.
Turns out, I'm "related" to Michael Kearney, indirectly through two marriages. My parents have met him.
(And my hair dresser's roommate's grandmother's bingo partner's boyfriend once sat next to Steve Jobs in an airport during a layover. That makes me special too.)
And what you clearly missed from the post you're responding to is that the clock rates that you can get from serial are so much higher than what you can do with parallel that it more than offsets the disadvantage of serialization.
There are two things that limit the speed of parallel interfaces. As the GP mentioned, one is signal skew. The clock rate has to be slow enough so that the receiver can sample all data lines at the same time and get them all within the data eye. The second is that the data lines are single-ended, meaning that there's only one wire per signal. The clock rate has to be slowed down to ensure that the signals have reached full on or full off at the other end and that they're noise free.
High-speed serial interfaces use DIFFERENTIAL SIGNALLING. The signal is transmitted over two wires that switch in antiphase. You decode them by comparing them. This has several beneficial effects. One is that noise affects them the same, so even if they're both offset by noise, they compare the same. The other is that now you don't have to wait as much on the effects of resistance, capacitance, and inductance. You can reliably decode the signal before the transitions are complete. (Look up "slew rate".)
So, using the same basic silicon technology, you can get a single differential pair to transmit data MUCH faster (in bytes/sec) than you can with parallel. It's interesting to see how technology transitioned from serial to parallel (wider means more bits per second), back to serial. The reason they didn't just stick with serial was that they just didn't have the technology to make the I/O drivers go that fast until recently.
IIRC, A 1x PCI Express channel is a single differential pair for data. (I think there's a side band channel and some other stuff.) This is just like DVI and SATA. 16x PCI Express is sixteen 1x channels. The trick here is that although data is interleaved across all 16 channels, those channels are not syncronized with each other. They are out of phase, and the the data is just put back into phase at the receiver.
I've mentioned this before, but some power users use software RAID, so it's relevant. Also, despite what some people tell you, properly configured purely software RAID under linux (mdadm, NOT dmraid) works very nicely, especially for RAID0 and RAID1.
Anyhow, it turns out that 9.04 is still unable to boot if the root volume is mdadm RAID1. As I understand it, this is because the initrd either doesn't load the mdadm driver or loads it too late. Non-root RAID volumes work fine.
This is a silly oversight that totally breaks 9.04 for anyone who wants to just put in two disks and run a mirrored configuration. And they still haven't fixed it. Note that Ubuntu has a long history of broken software RAID support while still claiming to support it, which is kinda dishonest, if you ask me. It's broken not because of a silly mistake but because they never bothered to test it before releasing.
Ok, so IBM isn't a monopoly and the should compete in the market. But this seems like a really pissy thing to do after the Sun deal went sour. They're just trying to be jerks about this.
This is a graphics card DEVELOPMENT PLATFORM. That implies a few things:
(1) This is a proving ground for designs that could be turned into a fast ASIC. (2) Graphics is only one of countless things you could use this for. How about using this as a basis for cryptographic offload, or high-end audio, or wifi?
Oh, and one other thing: Even in DOS, a lot of program (including the DOS command shell) completely bypass the int 0x10 API for drawing text, instead opting to write directly to the text buffer in the hardware. My host-only software solution was to hang off the 18hz system timer interrupt and just redraw the screen every interrupt. Of course, that goes away too when you switch to 32-bit protected mode.
[Architect of this VGA implemention writing] I have tried implementing VGA text mode emulation where the conversion from text (2 bytes per character) to graphics is done on the host CPU. This is doable for DOS apps, but as soon as the system enters protected mode, the interrupt table is blown away, and your console stops. The only solution is to do VGA text (emulation or not) in hardware, because the host (BIOS, Linux, etc.) expects the hardware to have a certain interface and do certain things. We chose to use a microcontroller for this purpose to minimize the over-all impact on the design.
Actually, I'm sure that many people can say they've come up with something like this. What's cool is that these people are commercializing it. My project idea involved putting small B/W LCD displays inside an array of physical push-buttons. Along with another larger fixed display, we had a fairly flexible device whose buttons could mean anything.
If free drivers are really a concern to you, you might consider helping out with a project that is working to develop a graphics card that itself is open source.
http://www.opengraphics.org
Consider making a donation to help out developers:
Trademarks aren't data or information. They're a name, and names are very important in human society. Moreover, names and trademarks do not infringe anyone's rights in the way that proprietary software does. I see no problem with a trademark holder telling you that you cannot use their name in an infringing way. It's not right for someone to go around using the name "Red Hat" or a derivative. If Wikimedia doesn't want their trademark infringed, they should have that right. Nothing stops you from using their Free licensed information. Creativity isn't being stifled. Technological development isn't being held back.
I installed the release candidate when it came out. Since I wanted to use it on a server, I decided to use software RAID. That was a nightmare all-around. First, their disk partitioner (the one that comes with ubuntu server anyhow) is junk. It gets horribly confused when there are existing partitions and won't let you remove them. I had to use dd to manually wipe sections of the disk before I could get the disk partitioned properly. After that, the installation went smoothly... until I it was time to reboot.
The machine wouldn't boot because it couldn't find the md devices. Some googling revealed some commands I could type into the busybox to get it to boot. But of course, I would have to enter these commands every time I booted.
And then the RAID array came up degraded. I did a lot of googling to find out how to rebuild the array, but every tutorial I found (including Ubuntu ones) mentoned device names (like/dev/sda1) that no longer exist in Ubuntu, so there was no way I could figure out the correct device node to use to re-add the second disk. And of course, it seems that it never occurred to Ubuntu developers to provide tools for this or do anything automatic.
So much for being a user-friendly distro. I actually had to go back to using Gentoo just so I could get what, to me, are basic things working. At least Gentoo documents stuff do you can fix it yourself. Ubuntu figures everything WILL be automatic, so they don't document it, so you're SoL when it doesn't work.
Software RAID has always taken a back seat for Ubuntu. I know one developer who has put a lot of work into it, but apparently his efforts have been thwarted by others. If you want a really basic desktop, Ubuntu is fine. But don't even consider it for anything server-related. The fact that they even offer a server edition makes them out to be dishonest.
Oh, and BTW, I went back and looked at the bug reports relating to software RAID not working. As far as I can tell, they havn't done anything to fix this between the RC and the release.
My wife got a Masters degree from a major university. But she got it remotely, doing classes via the internet. She was exposed to the same course material as anyone taking classes in person. But one critical thing that was missing was the interpersonal factor.
Doing it online, you miss the personal contact and information sharing with fellow students. Also, in grad school, it's important to make a good impression on professors, because you'll need references; that was lost too.
In my own experience, also as a grad student, I find that having set class times to go to keeps me organized and on-task. For those "project" classes where we mostly just worked in small groups, I just didn't feel like I learned as much, enjoyed it as much, or benefitted from a real-time sharing of ideas. As geeks, we often don't think about things like body language, but for those of us who are tuned into it, it too is a major component of the social aspects of communication and learning.
I'm also a telecommuter. Without weekly meetings and otherwise general in-person interaction, I quickly fell out of touch with day-to-day activities. I now rely too heavily on others to tell me what's going on and assign me work to do.
If formal classes "go away", you'll find some people who thrive, while you'll find others who flounder, finding it much harder to stay focused. Given that MOST people going to college are quite social, I suspect that organized in-person classes are here to stay. They may change shape somewhat, but there will always be classrooms with live lectures, at least until humans evolve into something quite different from how they are now.
I don't think classrooms are an accident that technology will erase. They're analogous to neandertals sitting around a fire, sharing the spoils of the hunt, telling their children about the way it's done, and discussing what they're going to do the next day. The clan getting together for dinner, as it were. Organizing into social groups is human nature. As such, everything we do starts with figuring out how and where the people are going to meet. The US Congress will stop in-person sessions long before classes do.
Slashdot Mirror
And I'm referring to the article and the guy who decided to do this "experiment."
(1) This software is designed for you to be able to license content licensable by AP. It assumes that the content you want to license was gotten from AP, because only people who want to license from AP will fill out the form to license text from AP. DUH.
(2) Technically, It may be prohibitive for them to write their software to sift through terabytes of text for every license request to ensure that the text you asked to license is (a) in there and (b) not licensed from someone else or public domain.
(3) It's fucking easy for anyone to guess that their software would behave this way.
(4) It's fucking unnecessary for it to be "smart" enough to second-guess stupid people who try to do this.
(5) Anyone knowingly trying to license from AP something not owned by AP is being a jerk.
(6) Anyone unknowingly trying to license from AP something not owned by AP is an idiot because they didn't do their homework.
(7) If you're a competent journalist, you'll know your rights, research the origin of content you want to quote, and attribute and license it appropriately. That's YOUR responsibility.
We all like a laugh at the expense of the **AA or the various media outlets. We are righteously outraged when they knowingly claim ownership of content that is not theirs. We are righteously outraged when they issue take-down notices for things that are obviously fair use.
But this is just fucking useless and stupid. I'm sorry. I'm pissed about this guy wasting our time with this, and some of you jackasses are swallowing it hook, line, and sinker. What pisses me off is that organizations like the AP are doing all kinds of things that REALLY ARE unfair, and yet some dick decides he's going to start picking on a non-issue. Now WE'RE the jackasses who are harrassing the AP. Two wrongs don't make a right.
I see nothing wrong with this service. I think it's perfectly sensible. The AP is just trying to save themselves and everyone else time and money by developing a convenient automated system for licensing content. Doing it this way saves on all sorts of legal, technical, and manpower costs. If they were to put in the effort to develop a system that would second-guess idiots who try to license content that's not actually under AP's control, it would cost ass-loads of money, and the service would be slow as shit. It wouldn't be worth the expense, so they'd shut it down and return to the old, expensive, manual way of doing things. You can argue about whether or not it's right for them to charge AT ALL for content. But that's a separate issue. Courts will uphold that they DO have a right to license at least SOME of their content, and this automated system is helping them keep volumes up and individual license fees down.
If you really want to catch the AP with their pants down, you'll have to find a situation (and I'm sure there are many) where the AP have explicitly (with a human involved) claimed ownership of content that wasn't theirs. In the case discussed in this article, I don't think it would stand up in any court that their automated system was being malicious, because the fact that they charged for a license doesn't necessarily imply ownership. Nowhere in any of this is any intent to defraud or deceive. The only way for you to be tricked by this system is to trick yourself.
Back when I was young (I graduated highschool in 1991), I recall people who migrated from WordPerfect to Word complaining about the missing "reveal codes" option. I looked into this, and this is what my friends with Ph.D.s at the time told me: Word didn't have "reveal codes" because it didn't have codes.
Let me step back and explain this a little better. Word Perfect used in-line codes to indicate formatting. There was an "italics on" code and there was an "italics off" command. It's not quite like HTML or XML, because it wasn't hierarchical. A document was a linear stream of bytes, and the word processor displayed the formatting by traversing the bytes to figure it out. On the processors of the day (386's), this had some major performance disadvantages, when the program had to scan back thousands of bytes just to figure out what the correct formatting was for what was being displayed on the screen. This was okay for the DOS version (can't see most of the formatting, so don't need to look for it), but it became a major liability for the Windows version. It was also a liability because documents that had been edited and edited tended to crud up with lots of superflous codes that WP simply didn't have the smarts to clean up. The only "advantage" was that you could reveal the codes, and that was only an advantage because people got used to it, and they got used to it because WP became problematic to use if you didn't reveal the codes to clean up problems.
Word did things differently. We all like to complain about Microsoft's behavior, and we like to complain about how crufty their software is. But now and then, their engineers (who are people like anyone else) did manage to do something that had intelligence behind it. Mind you, sometimes something has intelligence simply because someone thought about it and made an engineering decision. I'm not trying to claim that this was necessarily BETTER. Anyhow, Word didn't have reveal codes because it didn't have codes, per se, to reveal. Not in-line anyhow. Word was object-oriented. Word documents contained data structures that themselves indicated formatting and contained text. Paragraphs were objects. Sections were objects. Text within italics was inside an object. In a way, this is neither here nor there compared to reveal codes, but it made a practical difference in that when Word needed to determine the formatting of an object, rather than scanning back to the beginning of the file (which WP didn't always have to do but did sometimes which made it slow), Word worked its way up the object hierarchy, a much more efficient process. This also had advantages in that the object tree could be optimized to contain the formatting that was actually there. In WP, if you un-italicized a sentence that had been italicized, it wouldn't necessarily remove the old codes, instead inserting extra codes so that you got on's followed immediately by off's. Word would just delete the object.
So, to summarize, the reason Word didn't have reveal codes was that there were no in-line codes to reveal. Word's equivalent would have been some way to display the object hierarchy, which wouldn't necessarily have been intuitively useful to users. And of course, it would have been silly to emulate codes just to imitate a "feature" of WP that only existed in the first place because WP didn't automatically manage its codes properly.
We like to attribute scientific foresight to sci-fi. And indeed, sci-fi has been the inspiration for many a scientist or engineer. Star Trek writers probably picked aluminum because of its high strength and low density. But unless there was some kind of prior scientific advancement that they knew about, the choice of aluminum amounted to little more than a convenient plot device. It doesn't seem implausible to me that it would foreshadow the development of SOME kind of metal or other non-SiO2 material that's light, strong, and transparent. But not necessarily aluminum.
My office-mates were discussing this. Apparently, some professor on the radio was complaining that kids would be unable to read older documents written in cursive. And that seemed the best argument he could come up with.
Honestly, I think that's silly. They don't teach gothic script (Sütterlinschrift or Fraktur) in Germany anymore. Those who need it just LEARN it. This is no different from learning to read Chinese, Hebrew or particularly Arabic. They're not THAT hard to learn when you need them.
But maybe that's just me. Do most people have a really hard time learning to interpret a foreign script? It seems to me that it should be particularly easy if it's just a "code" for the same letters and words you use in your native language.
I wonder how well the space agencies are taking into consideration human need for intimate contact. NASA crews at least get some scheduled parts of the day when they get privacy and are disconnected from health monitors. But this is for relatively short missions. Unsatisfied sexual needs can create tension, especially in situations of extreme isolation with no way out. Perhaps only sexually active couples should be sent on these trips?
With respect to actually knowing things and getting work done, it's better to understand computer science concepts in general then to know obscure details of one or a few particular languages. But keep in mind that the people who hire progammers are not themselves programmers. Head hunters and HR departments are generally non-technical and will not understand (or care) if you talk about these concepts. The fact is, unless you're getting a Ph.D., you have to tailor your resume to be consumed by a half-broken computer program with multiple choice check boxes, and that is how you're going to be judged. It doesn't matter if you know C++, Java, Lisp, Ruby, PHP5, Python, Haskell, and assembly for 10 different processors. If the job calls for C#, and you don't claim to know it, you won't even be considered. If you DO claim to know it, it's hit or miss whether or not you'll be tested on it. I've actually gotten less than perfect scores on C and C++ tests due to things that I KNOW are wrong with the test, so you're up against that too. As much as you have to know computer science, you also have to know how to game the system.
Here's what has worked for me:
(1) Know your concepts well. Know how to take any program and implement it in pseudocode. Learn how to do it in structured, object oriented, and functional languages. This will help you actually get the job done and get raises and stuff once you're hired.
(2) Find out what languages are hot and read at least one small tutorial on each, and write programs in each of at least, say, 1000 lines. Whatever it takes to get used to the basics. Also, read the wikipedia articles and google for "trick and tips" and "gochas" and other things that people praise or gripe about for those languages. Get your feet wet. This way, you can then claim that you have programmed in each of those languages, and you can answer some of the weirder questions someone might ask.
(3) Familiarize yourself with different types programming environments or platforms. Program something on Windows. Program something on a Mac (if you can get your hands on one easily enough). Program something on Linux. Program something on an embedded system or at least a language used on them.
You have to lie judiciously. For instance, my background included assembly for several processors, C, C++, for modern system as well as things like 6502's, along with a fair amount of chip design experience. Could I do embedded systems programming? Duh. But at this one point, I didn't have any ACTUAL embedded systems programming experience. So I was honest and explained that while I hadn't, I clearly had enough related experience that was easy. I didn't get the job. What I should have done was done a bit of reading on the subject to ensure that I know the names of a few embedded processors and bluffed my way through. The problem was that the head hunter wasn't technical enough to understand my explanation, so all he knew to do was check "NO" on the embedded systems experience. He can comment in there about the chip design, but the HR people he would hand this off to wouldn't know what do do with it.
To summarize, to prepare yourself for working in the IT industry, you have to learn programming, but more importantly, you have to learn how to translate what you know into the language of the nontechnical people who do the hiring. This requires a kind of intelligence (subterfuge, to a limited extent) that many technophiles are not very skilled at. If you don't you give away control to the people who understand the art of deception better than they understand technical things. But those are the ones who rule the world. Politicians succeed not by knowing things but by knowing how to SPIN things.
There's an mdadm RAID5 driver, but I don't think dmraid does anything other than RAID0 and RAID1.
For RAID1, there's an optimization that one can make, which is to distribute reads between the two mirrors. For RAID0 and RAID5, there's only one place that any disk block can be read from, so there's no room for optization (or lack thereof).
See my other posts on this article. The Linux kernel has two separate software RAID drivers. "dmraid" interfaces with BIOS-based software RAID, while "mdraid" is Linux's native software RAID. There are two separate drivers. Unless you're trying to maintain compatibility with Windows using BIOS RAID, then dmraid is the wrong one to use. Use mdraid.
You are misinformed. dmraid (fakeraid) and mdraid (Linux native software RAID) _should_ use the same driver. The only real difference is where the metadata is stored (dmraid asks the BIOS, mdraid handles it in a more open way). Once the metadata is loaded, they're both just the OS accessing regular disks. What's the difference?
But in fact, there are two separate drivers. I discovered this by accidentally choosing the wrong one. When I discovered the difference (random reads are faster on mdraid but not dmraid), I asked on LKML and got no response. So I went digging through the kernel source myself and discovered the two separate drivers. The mdraid does a simple round-robbin. Whenever the next request is not contiguous from the last one (block number plus one), it switches to the other disk. No more intelligence than that. The dmraid, on the other hand, directs all reads to just the primary disk.
Heh. Linux software RAID doesn't do jack. I've looked at the source code. The mdadm RAID1 driver just alternates drives for reads whenever the requests are not contiguous. That is all. Nothing more. There's no intelligence in there. No keeping track of head positions, no attempts to discover or infer physical drive geometry. Nothing. Just a simple round-robbin. It just so happens that for MOST things that involve random access, the effective throughput is nearly doubled. More intelligence wouldn't actually buy you much in the general case, so why bother?
Also, the dmraid (fakeraid) RAID1 driver only does reads from one disk. I made the mistake of using dmraid instead of mdraid, only to discover through performance tests and iostat that there are basically two software RAID drivers that CLAIM to do identical things but in fact do not.
There goes our ability to suck down terabyte after terabyte of really lousy UUencoded porn.
(I find it more erotic in base64.)
Turns out, I'm "related" to Michael Kearney, indirectly through two marriages. My parents have met him.
(And my hair dresser's roommate's grandmother's bingo partner's boyfriend once sat next to Steve Jobs in an airport during a layover. That makes me special too.)
And what you clearly missed from the post you're responding to is that the clock rates that you can get from serial are so much higher than what you can do with parallel that it more than offsets the disadvantage of serialization.
There are two things that limit the speed of parallel interfaces. As the GP mentioned, one is signal skew. The clock rate has to be slow enough so that the receiver can sample all data lines at the same time and get them all within the data eye. The second is that the data lines are single-ended, meaning that there's only one wire per signal. The clock rate has to be slowed down to ensure that the signals have reached full on or full off at the other end and that they're noise free.
High-speed serial interfaces use DIFFERENTIAL SIGNALLING. The signal is transmitted over two wires that switch in antiphase. You decode them by comparing them. This has several beneficial effects. One is that noise affects them the same, so even if they're both offset by noise, they compare the same. The other is that now you don't have to wait as much on the effects of resistance, capacitance, and inductance. You can reliably decode the signal before the transitions are complete. (Look up "slew rate".)
So, using the same basic silicon technology, you can get a single differential pair to transmit data MUCH faster (in bytes/sec) than you can with parallel. It's interesting to see how technology transitioned from serial to parallel (wider means more bits per second), back to serial. The reason they didn't just stick with serial was that they just didn't have the technology to make the I/O drivers go that fast until recently.
IIRC, A 1x PCI Express channel is a single differential pair for data. (I think there's a side band channel and some other stuff.) This is just like DVI and SATA. 16x PCI Express is sixteen 1x channels. The trick here is that although data is interleaved across all 16 channels, those channels are not syncronized with each other. They are out of phase, and the the data is just put back into phase at the receiver.
It worked fine for me in 8.04 too. But it worked on and off in earlier versions and broke again in 9.04.
Also, this guide is badly out of date. Ubuntu doesn't use device nodes like /dev/sda anymore. They're referenced by UUID.
I've mentioned this before, but some power users use software RAID, so it's relevant. Also, despite what some people tell you, properly configured purely software RAID under linux (mdadm, NOT dmraid) works very nicely, especially for RAID0 and RAID1.
Anyhow, it turns out that 9.04 is still unable to boot if the root volume is mdadm RAID1. As I understand it, this is because the initrd either doesn't load the mdadm driver or loads it too late. Non-root RAID volumes work fine.
This is a silly oversight that totally breaks 9.04 for anyone who wants to just put in two disks and run a mirrored configuration. And they still haven't fixed it. Note that Ubuntu has a long history of broken software RAID support while still claiming to support it, which is kinda dishonest, if you ask me. It's broken not because of a silly mistake but because they never bothered to test it before releasing.
Ok, so IBM isn't a monopoly and the should compete in the market. But this seems like a really pissy thing to do after the Sun deal went sour. They're just trying to be jerks about this.
This is a graphics card DEVELOPMENT PLATFORM. That implies a few things:
(1) This is a proving ground for designs that could be turned into a fast ASIC.
(2) Graphics is only one of countless things you could use this for. How about using this as a basis for cryptographic offload, or high-end audio, or wifi?
Oh, and one other thing: Even in DOS, a lot of program (including the DOS command shell) completely bypass the int 0x10 API for drawing text, instead opting to write directly to the text buffer in the hardware. My host-only software solution was to hang off the 18hz system timer interrupt and just redraw the screen every interrupt. Of course, that goes away too when you switch to 32-bit protected mode.
[Architect of this VGA implemention writing] I have tried implementing VGA text mode emulation where the conversion from text (2 bytes per character) to graphics is done on the host CPU. This is doable for DOS apps, but as soon as the system enters protected mode, the interrupt table is blown away, and your console stops. The only solution is to do VGA text (emulation or not) in hardware, because the host (BIOS, Linux, etc.) expects the hardware to have a certain interface and do certain things. We chose to use a microcontroller for this purpose to minimize the over-all impact on the design.
Yes.
The XP10 contains these parts:
- PCI
- Microcontroller that does VGA
- PROM interfaces
The S3 is mostly empty and contains these parts:
- Memory controller
- Video controller
- Room for a graphics engine
Actually, I'm sure that many people can say they've come up with something like this. What's cool is that these people are commercializing it. My project idea involved putting small B/W LCD displays inside an array of physical push-buttons. Along with another larger fixed display, we had a fairly flexible device whose buttons could mean anything.
If free drivers are really a concern to you, you might consider helping out with a project that is working to develop a graphics card that itself is open source.
http://www.opengraphics.org
Consider making a donation to help out developers:
http://linuxfund.org/projects/ogd1/
Trademarks aren't data or information. They're a name, and names are very important in human society. Moreover, names and trademarks do not infringe anyone's rights in the way that proprietary software does. I see no problem with a trademark holder telling you that you cannot use their name in an infringing way. It's not right for someone to go around using the name "Red Hat" or a derivative. If Wikimedia doesn't want their trademark infringed, they should have that right. Nothing stops you from using their Free licensed information. Creativity isn't being stifled. Technological development isn't being held back.
Forget the law. This is an ethical issue.
I installed the release candidate when it came out. Since I wanted to use it on a server, I decided to use software RAID. That was a nightmare all-around. First, their disk partitioner (the one that comes with ubuntu server anyhow) is junk. It gets horribly confused when there are existing partitions and won't let you remove them. I had to use dd to manually wipe sections of the disk before I could get the disk partitioned properly. After that, the installation went smoothly... until I it was time to reboot.
The machine wouldn't boot because it couldn't find the md devices. Some googling revealed some commands I could type into the busybox to get it to boot. But of course, I would have to enter these commands every time I booted.
And then the RAID array came up degraded. I did a lot of googling to find out how to rebuild the array, but every tutorial I found (including Ubuntu ones) mentoned device names (like /dev/sda1) that no longer exist in Ubuntu, so there was no way I could figure out the correct device node to use to re-add the second disk. And of course, it seems that it never occurred to Ubuntu developers to provide tools for this or do anything automatic.
So much for being a user-friendly distro. I actually had to go back to using Gentoo just so I could get what, to me, are basic things working. At least Gentoo documents stuff do you can fix it yourself. Ubuntu figures everything WILL be automatic, so they don't document it, so you're SoL when it doesn't work.
Software RAID has always taken a back seat for Ubuntu. I know one developer who has put a lot of work into it, but apparently his efforts have been thwarted by others. If you want a really basic desktop, Ubuntu is fine. But don't even consider it for anything server-related. The fact that they even offer a server edition makes them out to be dishonest.
Oh, and BTW, I went back and looked at the bug reports relating to software RAID not working. As far as I can tell, they havn't done anything to fix this between the RC and the release.
My wife got a Masters degree from a major university. But she got it remotely, doing classes via the internet. She was exposed to the same course material as anyone taking classes in person. But one critical thing that was missing was the interpersonal factor.
Doing it online, you miss the personal contact and information sharing with fellow students. Also, in grad school, it's important to make a good impression on professors, because you'll need references; that was lost too.
In my own experience, also as a grad student, I find that having set class times to go to keeps me organized and on-task. For those "project" classes where we mostly just worked in small groups, I just didn't feel like I learned as much, enjoyed it as much, or benefitted from a real-time sharing of ideas. As geeks, we often don't think about things like body language, but for those of us who are tuned into it, it too is a major component of the social aspects of communication and learning.
I'm also a telecommuter. Without weekly meetings and otherwise general in-person interaction, I quickly fell out of touch with day-to-day activities. I now rely too heavily on others to tell me what's going on and assign me work to do.
If formal classes "go away", you'll find some people who thrive, while you'll find others who flounder, finding it much harder to stay focused. Given that MOST people going to college are quite social, I suspect that organized in-person classes are here to stay. They may change shape somewhat, but there will always be classrooms with live lectures, at least until humans evolve into something quite different from how they are now.
I don't think classrooms are an accident that technology will erase. They're analogous to neandertals sitting around a fire, sharing the spoils of the hunt, telling their children about the way it's done, and discussing what they're going to do the next day. The clan getting together for dinner, as it were. Organizing into social groups is human nature. As such, everything we do starts with figuring out how and where the people are going to meet. The US Congress will stop in-person sessions long before classes do.