You're overreaching there. The PET was released in 1977, the IBM PC in 1981. That's 4 years, not a decade. And both were marketed to business primarily, not consumers.
I'm skeptical that the PET was originally marketed to businesses with its funky chicklet keyboard. It was sold mostly to hobbyists, and perhaps schools. When Commodore changed the keyboard to a real keyboard, they rebranded it the CBM and then marketed it to businesses.
I own an iPhone 4. In general, the reception is worlds better than it was for my previous iPhone (a 3G).
It is indeed the case that if I hold the phone the way that I might prefer, the reception is only as good as it was on my previous iPhone. Are people really serious that Apple was somehow deficient in selling me a phone where I can now get much better reception than I used to?
The last time I saw this much negative commentary for an Apple product it was for the iPod mini. The majority of people on Slashdot claimed no one would want a music player that cost just as much and held fewer songs.
By this metric, the iPad will be a rousing success.
It turns out that you can use Apple TVs to drive AirPort Expresses, so you don't need to leave your computer on all the time.
Also an iPhone or an iPod Touch can act as a remote control for an Apple TV. The interface is the same as it is for an iPod Touch's or iPhone's iPod interface. I.e., excellent.
But... it requires iTunes, which means I have to run my noisy and power-hungry computer. Next best thing would be to snap up an old Mac Mini and stick iTunes on that, I suppose...
That's what I do, but my only computer at home is a Mac mini, so that was an easy decision! The mini draws very little power.
One could also buy an Apple TV and hack it to run a full version of OS X, if you want a a really cheap, small, quiet Mac (~$200). This is not for the feint of heart though.
Also, I assume that iTunes under Windoze can talk to Airport Expresses and Apple TVs (Apple TVs can also act as iTunes remote speakers), which means that your "power-hungry computer" could be a netbook running XP.
They were successful in creating an effective lightsaber in that it had a definite end point and would cut through anything, but when they attempted to cross swords, they just passed through one another... and then one of the people cut through the other one with the lightsaber he had. You can probably find it on youtube or on theforce.net somewhere...
Why in the world would IBM kill Java? IBM loves Java and probably does a lot more Java business than Sun.
Not that I'm a huge fan of IBM. I'd much prefer to see Sun remain Sun, while also figuring out how to stay in business.
Regarding Google buying Sun, I don't think Google has the slightest interest in Solaris, which would be a shame, since Solaris 10 is pretty darn sweet.
Please. You get nearly twice as much PC for the same price as a Mac Mini
So? Price isn't the metric for everything. Sometimes you're willing to pay more to get something smaller and quieter and sometimes not.
There's absolutely no more room on my desk at work, so when I need another computer, I get a mini. When I want a computer for my stereo cabinet at home, I get a mini.
I get it -- YOU don't want to pay extra for a mini. Bully for you. Don't speak for all Unix fans.
I've never in my life, met a UNIX nerd who gave the slightest damn what his computer looked like, assuming you could even see the damn thing in the chaos.
Maybe you just don't get out much.
That's funny, so am I. I even get paid to be one. And I find the idea that OS X is as "UNIXy" as my Solaris, FreeBSD - and even Linux - servers, to be laughable.
Being as Unixy as FreeBSD is largely irrelevant to me, since nobody anywhere around me uses FreeBSD even though everyone uses Unix or Linux. I'm sure I'd love to use FreeBSD, as I cut my teeth on BSD 4.2. But I've grown tired at tilting at windmills.
If you ask me, Solaris and Linux are just as far away from being "real" Unix as OS X because they've strayed just as far from BSD 4.3 as OS X has. I'm equally comfortable on any one of them, because they all made different bone-headed mistakes. They're still all Unix to me, despite their warts.
OS X is firstly, foremost, primarily and overpoweringly, a desktop platform that just happens to look a bit like a typical UNIX if you scratch the surface.
Assertions such as this are not just laughable, they're inane. OS X is Unix when you scratch the surface. The fact that most people don't scratch the surface is irrelevant, as all the people I work with, do go past the surface because they're Unix people, just like me.
Btw, I'm paid to be a Unix wizard too. I've even been the head system administrator at a large research lab with hundreds of Unix workstations and servers. I've (among many other "wizardly" things) written sendmail.cf files from scratch, implemented scripts that would upgrade hundreds of BSD workstations at a time for Project Athena, maintained lab-wide DNS, NFS, timesharing, and mail servers, written multiple backup systems, including one based on rsync and rotating snapshots, helped spec out and purchase HPC clusters and parallel filesystems, have purchased a few million dollars worth of Unix and network equipment, and have given talks at "large installation" Unix sysadmin user groups.
Please don't deign to assert to me that ALL Unix experts have the same preferences as you, because that's a blatant falsehood.
Please. You're advocating OS X and trying to make an argument about _performance_ ? And you then have the temerity to question _my_ credibility ?
Please -- you're recommending a system that's supposed to make Unix-lovers happy that can't even reliably handle ^C and ^Z. It surely doesn't make me happy to not have these things work. A true Unix lover, if relegated to Windows, would run BSD or Solaris in a VM. Personally, I'd rather just have one integrated system that works well.
Regarding performance, common tasks take 40 times longer under Cygwin than under Linux, for instance. OS X has no such performance issues. Yes, there may be a 20% penalty on some things due to going through a microkernel. I don't lose any sleep over this -- my Macs are zippy enough, and some things it does significantly faster than other Unixes. (E.g., SQLite does some operations 5 times faster under OS X than under Ubuntu.) The biggest performance issue for me is using MH with a million emails in single folder under HFS. HFS doesn't like this. BFD -- I format an OS X disk partition with UFS and run MH on that.
Maybe it's true in your little world, but it's not true in mine.
Which makes no sense. Why would you pay the Apple tax for a pretty face on X11, xterm and emacs when you can get the same thing from a Linux machine (or even an OpenSolaris PC, if you're a traditionalist) for probably half the price ?
Many reasons, including:
(1) To get a Unix machine that works out of the box without a lot of fiddling. That works with your network card, and your display card. That works with a 30-inch monitor without endless hacking on the XF86Config file.
We had an employee who insisted on a Linux notebook computer. It never worked for him. He couldn't get the display driver to work with whatever weird video card Lenovo was shipping that week.
(2) To be able to run more polished or popular commercial apps when you want to, even if that's not the main thing that you do.
(3) Mac Books have excellent industrial design.
(4) Mac minis are small and quiet and not much more expensive than inferior imitators.
(5) Etc., etc., etc.
There are many excellent reasons to use OS X. That your primary interest is a familiar and typical UNIX-like environment, but with a pretty face, is _not_ one of them, because the UNIX aspect of OS X is neither familiar, nor typical, once you move past trivial usage (stuff even Cygwin does just as well).
You haven't a clue. I'm a Unix wizard. OS X's Unix is completely familiar and typical to me. Sure I have to use fink or Ports to make it so. So what? They're no better or worse than the package managers on any other Unix/Linux.
Regarding Cygwin -- you're nuts. It can't handle signals properly and does forks incredibly slowly. Also the NT filesystem really bites when you're looking to just be happy with Unix.
Regarding the Apple tax, my precious time is worth oh so much more than a few bucks. You can be penny wise and pound foolish if you want. Many people chose otherwise. Or, if you have fun endlessly fiddling, feel free. I used to have fun with that sort of crap too. Now I prefer to get other stuff done.
You can have whatever opinion you want, but your facts are wrong.
Of all the bits of OS X that are actually interesting and of value to users, "it's a UNIX" is a long, long, long way down the list.
To which users? For the majority of users surely you are correct, but we weren't talking about the majority of computer users-- we are talking about Unix users.
OS X is cleaning up in the university worlds where I live, because most of the Unix nerds, such as myself, are perfectly happy with OS X, which I use mostly as a pretty front-end to X11, xterm, and emacs.
I was hoping that it would be an easy way to write Ajax based pages with out needing to dig to deep in to javaScript and all it's cross browser issues. But I guess not.
I'm confused by your comment about GWT. The goal of GWT is precisely so that you can make Ajax-based sites without every writing a single line of Javascript or worry about cross-browser issues. You just write all your code in Java using their widget set, and their compiler does all the rest for you automatically.
Of course, this only works for browsers that GWT supports.
A lot of Google code is written in Python. It's not a big secret. In fact, I can't think of any Google product that uses Java for its backend, Java is just plain-old too slow for Google.
This is just nonsense. I'm a huge Python fan, but for most purposes Java is much faster, and on the server, Java is typically the fastest language around, since the JIT compiler is constantly optimizing the hell out of the "hotspots".
I just don't get the argument that this is a bad idea. If you ask me, it's the best idea ever. Top level domains were a stupid idea to begin with. There's no need, for example, for there to be distinct sites "ibm.com", "ibm.net", and "ibm.org". There should just be "ibm".
As I understand this proposal, by completely opening up the top level domains, it effectively eliminates top level domains, and makes things the way they should have been to begin with.
BodyBilt makes the most ergonomic chairs I've ever seen. They're rather pricey, but I developed terrible carpal tunnel syndrome and back problems more than a decade ago. I thought I was going to have to go on disability, but a BodyBilt chair with "linear tracking arms" and a Kinesis Contour Keyboard completely saved my ass. They brought about a near complete recovery, and I've had few problems since.
If it were up to me, these chairs and keyboards would be legally mandated equipment!
Yes, the features that you want have been among the promised improvements to ZFS for a couple of years now. RAIDZ2 was one of the promised features, and it became reality a while back. Some of the other improvements have not yet materialized.
Despite any of this, however, ZFS is not worse in any of these regards than any other common filesystem/volume manager/RAID device.
Regarding adding disks to a pool, ZFS doesn't yet support adding additional disks to a RAIDZ, but it does support replacing a disk in a RAIDZ with a larger disk (and the space will be used if it can be, modulo redundancy requirements) and it does support expanding a pool by adding new RAIDZs, mirrored pairs, or individual disks to the pool. Individual disks, however, add single points of failure, so it will be very nice when you can add a disk onto a RAIDZ, and it will also be nice when you can remove a disk from a pool if you have enough space elsewhere in the pool to absorb the loss. These two feature-requests are still on the ZFS to-do list.
With ZFS, a pool is a collection of "vdevs", and you can add new vdevs to the pool at any time to increase the capacity of the pool. A vdev is either a RAIDZ (which is kind of like RAID5), a RAIDZ2 (which is kind of like RAID6), a mirror (which is kind of like RAID1), or a bare disk. The pool is then is kind of like a RAID0 over all the vdevs.
I believe your evaluation to be incorrect on several levels. Firstly, the issue you point out is true for RAID-anything, as the filesystem has to be able to survive the loss of one of the disks for RAIDZ. RAID5 is no different in this regard.
Secondly, with RAIDZ (or RAID5) and 4x500GB, you wouldn't end up with 2TB of disk space -- you'd end up with 1.5TB due to the overhead of the parity data.
Thirdly, you don't have to replace all of the disk drives with RAIDZ to increase the amount of disk space dramatically. You seem to be thinking of RAID5, not RAIDZ. With RAIDZ replacing one of your 500GB disk drives with a new 2TB disk drive would indeed still leave you with only 1.5TB of disk space, due to the requirement for redundancy, but if you bought a pair of 2TB disk drives to replace two of your 500GB disk drives, you would increase your disk capacity from 1.5TB to 3TB, and if you just added the pair of 2TB disk drives to the pool as a mirror, as opposed to replacing existing drives, then you'd increase your disk capacity to 3.5TB.
Fourth, no one is forcing you to use redundancy with ZFS if you don't want to suffer the redundancy/reliability overhead. You can add non-redundant disk drives to a ZFS pool.
If you want extra reliability, you have to pay for it somehow.
I don't really have a strong opinion on this issue one way or the other, but I do know that more than a decade ago I read an article in The Atlantic Monthly, which is a rather Liberal journal, that there seemed to be no evidence that plutonium was nearly as dangerous as previously thought.
According to the article, all the research claiming to prove that plutonium is terribly carcinogenic had been done on dogs. That particular evidence seemed to stand up. But for whatever reasons, this phenomenon didn't seem to translate to humans. According to the article, the mortality rate among people who had worked in plutonium factories decades ago, was no higher than average. Some of these workers were so radioactive that to this day they would set off alarms in buildings that have radiation detectors, just by walking into them.
Wow, it's pretty hard to believe that only 3.3% of programmers can code a simple logic problem correctly when given pencil, paper, no pressure, and a bit of time. It's actually a well-known phenomenon of cognitive psychology, though, that there are some very simple logic problems that even people with PhDs will almost always get wrong if rushed. Even people with PhDs in logic. (But probably not people with PhDs in Cognitive Psychology, because they will know of these puzzles already, and be extra careful.)
I'm not sure what this implies about the quality of the programming applicant pool, though. Just because the vast majority of highly intelligent and skilled people are known to typically fall into the simplest of logic traps, doesn't mean that they don't actually perform well in the real world. In the real world, people have time to write a little test case, drink a cup of coffee, and get the code working before checking it into revision control. The sign of a good programmer is not to get things correctly on the first attempt, but, rather to know to test thoroughly any code that is likely to be tricky at all. (And even code that is not.)
Consequently, I'm pretty suspicious of interviewing via tests. That only assures you that you'll get someone who tests well. Does that mean that they can successfully work on a large project with a code base of millions of lines of code? Does it mean that they can ever understand all the subtle nuances of a programming language like C++? I'd prefer to evaluate people based on what they've accomplished and perhaps some reviewing of some tricky or sophisticated code they've previously written and maybe some documentation samples. Those things are more representative of the skills required in the real world.
|>oug
P.S. A quick screening question I might ask a C++ programmer is if you define a copy constructor, what else must you typically also define? Or if you declare a class with any virtual methods, what is the one particular method that then must also be virtual. If someone doesn't know these standard C++ pitfalls in their sleep, they they will wreak havoc upon your code.
That's precisely what I did at an interview where I was asked to program at the interview, and I got the job. See another comment I made in this subthread for more details.
Unfortunately, companies such as Google don't take no for an answer when they ask you to code right at the whiteboard.
Actually, I do work at Harvard, on bio/medical technology. My code tends to be very robust.
I've worked fine under deadlines. In fact on software that had to work by the time a Delta II rocket was launched. Deadlines don't get any firmer than that. Deadline pressure is not the same as stage fright.
Anyone who has graduated from college with good grades should be able to work under deadline pressure. Working under stage fright pressure is something that you would work through in graduate school while teaching classes.
In the real world, however, I've never had to work under stage fright pressure, except to give talks. And in those situations, I'm very rehearsed.
(1) Stage fright is completely different from deadline stress. For instance, I've done interview coding, but I told the interviewers to give me pencil and paper and leave and come back in a while. I got the job, as I was the only one who could solve the problems at all, much less at a whiteboard. (2) Client B never gets a code release on that sort of time frame any place I've ever worked. Code released publicly has to go through code review, QA, etc. Internal clients, on the other hand, are generally reasonable people, and don't threaten to take away all their business if you can't solve the problem on some crazy deadline. (3) Most bugs are very much more complicated than could ever be fixed on such a tight schedule. (4) My code is generally much more robust than that of most programmers, so I'd likely save the company from that problem in the first place.
You do realize, don't you, that you are not testing what you think you're testing. You are testing someone's ability to program under extreme stress, not their ability to program.
I graduated from MIT, got A+'s in all my programming and software engineering classes, have written software that operates an X-ray space telescope, taught myself C++ in a week by reading Stroustrup from cover to cover, scored in the top 5% on the computer science and general GREs, and have derived the linear algebra of coordinate frame transformations on my own from first principles because I didn't happen to have a good textbook handy.
On the other hand, I'd likely fail your test. I'd be lucky to be able to remember my own name if you asked me to write it down on a whiteboard during an interview.
Slashdot Mirror
You're overreaching there. The PET was released in 1977, the IBM PC in 1981. That's 4 years, not a decade. And both were marketed to business primarily, not consumers.
I'm skeptical that the PET was originally marketed to businesses with its funky chicklet keyboard. It was sold mostly to hobbyists, and perhaps schools. When Commodore changed the keyboard to a real keyboard, they rebranded it the CBM and then marketed it to businesses.
|>ouglas
Wow, way to buy into all the marketing speak.
Wow, way to buy into the FUD!
I own an iPhone 4. In general, the reception is worlds better than it was for my previous iPhone (a 3G).
It is indeed the case that if I hold the phone the way that I might prefer, the reception is only as good as it was on my previous iPhone. Are people really serious that Apple was somehow deficient in selling me a phone where I can now get much better reception than I used to?
|>ouglas
The last time I saw this much negative commentary for an Apple product it was for the iPod mini. The majority of people on Slashdot claimed no one would want a music player that cost just as much and held fewer songs.
By this metric, the iPad will be a rousing success.
|>ouglas
It turns out that you can use Apple TVs to drive AirPort Expresses, so you don't need to leave your computer on all the time.
Also an iPhone or an iPod Touch can act as a remote control for an Apple TV. The interface is the same as it is for an iPod Touch's or iPhone's iPod interface. I.e., excellent.
|>ouglas
But... it requires iTunes, which means I have to run my noisy and power-hungry computer. Next best thing would be to snap up an old Mac Mini and stick iTunes on that, I suppose...
That's what I do, but my only computer at home is a Mac mini, so that was an easy decision! The mini draws very little power.
One could also buy an Apple TV and hack it to run a full version of OS X, if you want a a really cheap, small, quiet Mac (~$200). This is not for the feint of heart though.
Also, I assume that iTunes under Windoze can talk to Airport Expresses and Apple TVs (Apple TVs can also act as iTunes remote speakers), which means that your "power-hungry computer" could be a netbook running XP.
|>ouglas
They were successful in creating an effective lightsaber in that it had a definite end point and would cut through anything, but when they attempted to cross swords, they just passed through one another... and then one of the people cut through the other one with the lightsaber he had. You can probably find it on youtube or on theforce.net somewhere...
Indeed you can find it on YouTube. Here it is:
http://www.youtube.com/watch?v=dsZNiCSCLXw
|>ouglas
Why in the world would IBM kill Java? IBM loves Java and probably does a lot more Java business than Sun.
Not that I'm a huge fan of IBM. I'd much prefer to see Sun remain Sun, while also figuring out how to stay in business.
Regarding Google buying Sun, I don't think Google has the slightest interest in Solaris, which would be a shame, since Solaris 10 is pretty darn sweet.
|>ouglas
Please. You get nearly twice as much PC for the same price as a Mac Mini
So? Price isn't the metric for everything. Sometimes you're willing to pay more to get something smaller and quieter and sometimes not.
There's absolutely no more room on my desk at work, so when I need another computer, I get a mini. When I want a computer for my stereo cabinet at home, I get a mini.
I get it -- YOU don't want to pay extra for a mini. Bully for you. Don't speak for all Unix fans.
I've never in my life, met a UNIX nerd who gave the slightest damn what his computer looked like, assuming you could even see the damn thing in the chaos.
Maybe you just don't get out much.
That's funny, so am I. I even get paid to be one. And I find the idea that OS X is as "UNIXy" as my Solaris, FreeBSD - and even Linux - servers, to be laughable.
Being as Unixy as FreeBSD is largely irrelevant to me, since nobody anywhere around me uses FreeBSD even though everyone uses Unix or Linux. I'm sure I'd love to use FreeBSD, as I cut my teeth on BSD 4.2. But I've grown tired at tilting at windmills.
If you ask me, Solaris and Linux are just as far away from being "real" Unix as OS X because they've strayed just as far from BSD 4.3 as OS X has. I'm equally comfortable on any one of them, because they all made different bone-headed mistakes. They're still all Unix to me, despite their warts.
OS X is firstly, foremost, primarily and overpoweringly, a desktop platform that just happens to look a bit like a typical UNIX if you scratch the surface.
Assertions such as this are not just laughable, they're inane. OS X is Unix when you scratch the surface. The fact that most people don't scratch the surface is irrelevant, as all the people I work with, do go past the surface because they're Unix people, just like me.
Btw, I'm paid to be a Unix wizard too. I've even been the head system administrator at a large research lab with hundreds of Unix workstations and servers. I've (among many other "wizardly" things) written sendmail.cf files from scratch, implemented scripts that would upgrade hundreds of BSD workstations at a time for Project Athena, maintained lab-wide DNS, NFS, timesharing, and mail servers, written multiple backup systems, including one based on rsync and rotating snapshots, helped spec out and purchase HPC clusters and parallel filesystems, have purchased a few million dollars worth of Unix and network equipment, and have given talks at "large installation" Unix sysadmin user groups.
Please don't deign to assert to me that ALL Unix experts have the same preferences as you, because that's a blatant falsehood.
Please. You're advocating OS X and trying to make an argument about _performance_ ? And you then have the temerity to question _my_ credibility ?
Please -- you're recommending a system that's supposed to make Unix-lovers happy that can't even reliably handle ^C and ^Z. It surely doesn't make me happy to not have these things work. A true Unix lover, if relegated to Windows, would run BSD or Solaris in a VM. Personally, I'd rather just have one integrated system that works well.
Regarding performance, common tasks take 40 times longer under Cygwin than under Linux, for instance. OS X has no such performance issues. Yes, there may be a 20% penalty on some things due to going through a microkernel. I don't lose any sleep over this -- my Macs are zippy enough, and some things it does significantly faster than other Unixes. (E.g., SQLite does some operations 5 times faster under OS X than under Ubuntu.) The biggest performance issue for me is using MH with a million emails in single folder under HFS. HFS doesn't like this. BFD -- I format an OS X disk partition with UFS and run MH on that.
|>ouglas
If you call yourself one, chances are, you aren't.
You can deal in chances if you want. I'll deal in facts.
|>ouglas
People who want UNIX don't use OS X.
You couldn't be more wrong.
Maybe it's true in your little world, but it's not true in mine.
Which makes no sense. Why would you pay the Apple tax for a pretty face on X11, xterm and emacs when you can get the same thing from a Linux machine (or even an OpenSolaris PC, if you're a traditionalist) for probably half the price ?
Many reasons, including:
(1) To get a Unix machine that works out of the box without a lot of fiddling. That works with your network card, and your display card. That works with a 30-inch monitor without endless hacking on the XF86Config file.
We had an employee who insisted on a Linux notebook computer. It never worked for him. He couldn't get the display driver to work with whatever weird video card Lenovo was shipping that week.
(2) To be able to run more polished or popular commercial apps when you want to, even if that's not the main thing that you do.
(3) Mac Books have excellent industrial design.
(4) Mac minis are small and quiet and not much more expensive than inferior imitators.
(5) Etc., etc., etc.
There are many excellent reasons to use OS X. That your primary interest is a familiar and typical UNIX-like environment, but with a pretty face, is _not_ one of them, because the UNIX aspect of OS X is neither familiar, nor typical, once you move past trivial usage (stuff even Cygwin does just as well).
You haven't a clue. I'm a Unix wizard. OS X's Unix is completely familiar and typical to me. Sure I have to use fink or Ports to make it so. So what? They're no better or worse than the package managers on any other Unix/Linux.
Regarding Cygwin -- you're nuts. It can't handle signals properly and does forks incredibly slowly. Also the NT filesystem really bites when you're looking to just be happy with Unix.
Regarding the Apple tax, my precious time is worth oh so much more than a few bucks. You can be penny wise and pound foolish if you want. Many people chose otherwise. Or, if you have fun endlessly fiddling, feel free. I used to have fun with that sort of crap too. Now I prefer to get other stuff done.
You can have whatever opinion you want, but your facts are wrong.
|>ouglas
Of all the bits of OS X that are actually interesting and of value to users, "it's a UNIX" is a long, long, long way down the list.
To which users? For the majority of users surely you are correct, but we weren't talking about the majority of computer users-- we are talking about Unix users.
OS X is cleaning up in the university worlds where I live, because most of the Unix nerds, such as myself, are perfectly happy with OS X, which I use mostly as a pretty front-end to X11, xterm, and emacs.
I was hoping that it would be an easy way to write Ajax based pages with out needing to dig to deep in to javaScript and all it's cross browser issues. But I guess not.
I'm confused by your comment about GWT. The goal of GWT is precisely so that you can make Ajax-based sites without every writing a single line of Javascript or worry about cross-browser issues. You just write all your code in Java using their widget set, and their compiler does all the rest for you automatically.
Of course, this only works for browsers that GWT supports.
|>oug
A lot of Google code is written in Python. It's not a big secret. In fact, I can't think of any Google product that uses Java for its backend, Java is just plain-old too slow for Google.
This is just nonsense. I'm a huge Python fan, but for most purposes Java is much faster, and on the server, Java is typically the fastest language around, since the JIT compiler is constantly optimizing the hell out of the "hotspots".
|>oug
It sounds like you are thinking about applets. That is a LONG dead technology.
No it isn't. The "Processing" IDE/API breathes new life into applets. Here's a Java applet I wrote and it is quite nifty keen:
http://am.iic.harvard.edu/DendroStar
|>oug
I just don't get the argument that this is a bad idea. If you ask me, it's the best idea ever. Top level domains were a stupid idea to begin with. There's no need, for example, for there to be distinct sites "ibm.com", "ibm.net", and "ibm.org". There should just be "ibm".
As I understand this proposal, by completely opening up the top level domains, it effectively eliminates top level domains, and makes things the way they should have been to begin with.
|>oug
BodyBilt makes the most ergonomic chairs I've ever seen. They're rather pricey, but I developed terrible carpal tunnel syndrome and back problems more than a decade ago. I thought I was going to have to go on disability, but a BodyBilt chair with "linear tracking arms" and a Kinesis Contour Keyboard completely saved my ass. They brought about a near complete recovery, and I've had few problems since.
If it were up to me, these chairs and keyboards would be legally mandated equipment!
|>oug
Yes, the features that you want have been among the promised improvements to ZFS for a couple of years now. RAIDZ2 was one of the promised features, and it became reality a while back. Some of the other improvements have not yet materialized.
Despite any of this, however, ZFS is not worse in any of these regards than any other common filesystem/volume manager/RAID device.
Regarding adding disks to a pool, ZFS doesn't yet support adding additional disks to a RAIDZ, but it does support replacing a disk in a RAIDZ with a larger disk (and the space will be used if it can be, modulo redundancy requirements) and it does support expanding a pool by adding new RAIDZs, mirrored pairs, or individual disks to the pool. Individual disks, however, add single points of failure, so it will be very nice when you can add a disk onto a RAIDZ, and it will also be nice when you can remove a disk from a pool if you have enough space elsewhere in the pool to absorb the loss. These two feature-requests are still on the ZFS to-do list.
|>oug
With ZFS, a pool is a collection of "vdevs", and you can add new vdevs to the pool at any time to increase the capacity of the pool. A vdev is either a RAIDZ (which is kind of like RAID5), a RAIDZ2 (which is kind of like RAID6), a mirror (which is kind of like RAID1), or a bare disk. The pool is then is kind of like a RAID0 over all the vdevs.
|>oug
I believe your evaluation to be incorrect on several levels. Firstly, the issue you point out is true for RAID-anything, as the filesystem has to be able to survive the loss of one of the disks for RAIDZ. RAID5 is no different in this regard.
Secondly, with RAIDZ (or RAID5) and 4x500GB, you wouldn't end up with 2TB of disk space -- you'd end up with 1.5TB due to the overhead of the parity data.
Thirdly, you don't have to replace all of the disk drives with RAIDZ to increase the amount of disk space dramatically. You seem to be thinking of RAID5, not RAIDZ. With RAIDZ replacing one of your 500GB disk drives with a new 2TB disk drive would indeed still leave you with only 1.5TB of disk space, due to the requirement for redundancy, but if you bought a pair of 2TB disk drives to replace two of your 500GB disk drives, you would increase your disk capacity from 1.5TB to 3TB, and if you just added the pair of 2TB disk drives to the pool as a mirror, as opposed to replacing existing drives, then you'd increase your disk capacity to 3.5TB.
Fourth, no one is forcing you to use redundancy with ZFS if you don't want to suffer the redundancy/reliability overhead. You can add non-redundant disk drives to a ZFS pool.
If you want extra reliability, you have to pay for it somehow.
|>oug
I don't really have a strong opinion on this issue one way or the other, but I do know that more than a decade ago I read an article in The Atlantic Monthly, which is a rather Liberal journal, that there seemed to be no evidence that plutonium was nearly as dangerous as previously thought.
According to the article, all the research claiming to prove that plutonium is terribly carcinogenic had been done on dogs. That particular evidence seemed to stand up. But for whatever reasons, this phenomenon didn't seem to translate to humans. According to the article, the mortality rate among people who had worked in plutonium factories decades ago, was no higher than average. Some of these workers were so radioactive that to this day they would set off alarms in buildings that have radiation detectors, just by walking into them.
|>oug
Wow, it's pretty hard to believe that only 3.3% of programmers can code a simple logic problem correctly when given pencil, paper, no pressure, and a bit of time. It's actually a well-known phenomenon of cognitive psychology, though, that there are some very simple logic problems that even people with PhDs will almost always get wrong if rushed. Even people with PhDs in logic. (But probably not people with PhDs in Cognitive Psychology, because they will know of these puzzles already, and be extra careful.)
I'm not sure what this implies about the quality of the programming applicant pool, though. Just because the vast majority of highly intelligent and skilled people are known to typically fall into the simplest of logic traps, doesn't mean that they don't actually perform well in the real world. In the real world, people have time to write a little test case, drink a cup of coffee, and get the code working before checking it into revision control. The sign of a good programmer is not to get things correctly on the first attempt, but, rather to know to test thoroughly any code that is likely to be tricky at all. (And even code that is not.)
Consequently, I'm pretty suspicious of interviewing via tests. That only assures you that you'll get someone who tests well. Does that mean that they can successfully work on a large project with a code base of millions of lines of code? Does it mean that they can ever understand all the subtle nuances of a programming language like C++? I'd prefer to evaluate people based on what they've accomplished and perhaps some reviewing of some tricky or sophisticated code they've previously written and maybe some documentation samples. Those things are more representative of the skills required in the real world.
|>oug
P.S. A quick screening question I might ask a C++ programmer is if you define a copy constructor, what else must you typically also define? Or if you declare a class with any virtual methods, what is the one particular method that then must also be virtual. If someone doesn't know these standard C++ pitfalls in their sleep, they they will wreak havoc upon your code.
That's precisely what I did at an interview where I was asked to program at the interview, and I got the job. See another comment I made in this subthread for more details.
Unfortunately, companies such as Google don't take no for an answer when they ask you to code right at the whiteboard.
|>oug
Actually, I do work at Harvard, on bio/medical technology. My code tends to be very robust.
I've worked fine under deadlines. In fact on software that had to work by the time a Delta II rocket was launched. Deadlines don't get any firmer than that. Deadline pressure is not the same as stage fright.
Anyone who has graduated from college with good grades should be able to work under deadline pressure. Working under stage fright pressure is something that you would work through in graduate school while teaching classes.
In the real world, however, I've never had to work under stage fright pressure, except to give talks. And in those situations, I'm very rehearsed.
|>oug
(1) Stage fright is completely different from deadline stress. For instance, I've done interview coding, but I told the interviewers to give me pencil and paper and leave and come back in a while. I got the job, as I was the only one who could solve the problems at all, much less at a whiteboard. (2) Client B never gets a code release on that sort of time frame any place I've ever worked. Code released publicly has to go through code review, QA, etc. Internal clients, on the other hand, are generally reasonable people, and don't threaten to take away all their business if you can't solve the problem on some crazy deadline. (3) Most bugs are very much more complicated than could ever be fixed on such a tight schedule. (4) My code is generally much more robust than that of most programmers, so I'd likely save the company from that problem in the first place.
You do realize, don't you, that you are not testing what you think you're testing. You are testing someone's ability to program under extreme stress, not their ability to program.
I graduated from MIT, got A+'s in all my programming and software engineering classes, have written software that operates an X-ray space telescope, taught myself C++ in a week by reading Stroustrup from cover to cover, scored in the top 5% on the computer science and general GREs, and have derived the linear algebra of coordinate frame transformations on my own from first principles because I didn't happen to have a good textbook handy.
On the other hand, I'd likely fail your test. I'd be lucky to be able to remember my own name if you asked me to write it down on a whiteboard during an interview.
|>oug