Are you serious? A URL is an address, not necessarily something you can click on./. left off the HREF on purpose so that people wouldn't blindly click and get burned by the malicious code.
Sure you can, if you fix the application. Complex calculations are not just a single operation, they're a series of smaller operations that are strung together resulting in a more comples argorithm. What I'm suggesting, is to alter the application so that it saves checkpoints along the way.
This reminds me of the situation that I came across while working in the computer labs in college. After people lost their paper a few times, they learned to click Save every so often. That's much easier than some fancy hybernation system people are talking about.
If you're implying that HP would be in better shape now if they had kept the T&M divisions, I disagree. Agilent is in worse share than HP is right now.
You might think that Jouko Pynnonen was just catering to Microsoft when he's refusing to release more details about how to exploit the bug, but I think he's doing the right thing. The whole goal, after all, is to make things more secure. And getting Microsoft to fix the problem, while at the same time not giving crackers the full information they need is probably the closest thing to a good security situation we can have here (aside from not having the bug in the first place of course).
There's an excellent article by Bruce Schneier in his latest Crypto-Gram newsletter discussing the issue of Full Disclosure. I recommend taking a look at it.
I've had Speakeasy as my ISP for about 18 months now, and I've been absolutely delighted with them. Not only do they not mind if I run servers off of my DSL line, but the connection has been extremely reliable. The line has been down just a few hours (one day) in the last 18 months.
There are two unix machines named A and B that are on the same subnet. Describe to me, in as much detail as possible, what happens when I type "telnet B" from a terminal on machine A.
The "in as much detail as possible" is the key phrase here. The interviewer got to see an understanding of (or lack of) PATH, inetd, DNS, subnetting, TCP/IP, ethernet, etc.
That question, and the discussion we had afterward impressed me so much about the technical caliber of the manger, I took the job.
I worked for HP for 5 years as a unix admin for manufacturing sites. I was always part of an on-call rotation. HP has what it calls GAP. It stands for Guaranteed Availability Pay. Basically, instead of paying you only when you get called in (and encouraging unreliable systems), they would pay you for being on-call. There would be no extra pay beyond GAP for a call-in.
I though this made a lot of sense because you still got paid even if there were no problems. It really encouraged you to make the systems as reliable as possible.
For that kind of money, you can afford to get a big box and an industrial strength unix like HP-UX, Solaris, or AIX. Linux is great if you don't have a lot of money and aren't going to need lots of processors and lots of RAM, but I don't think it's a good fit for what you're talking about.
It's definitely better if you can have a technical manager, but from my experience, it's rare. For some reason, it seems that you never find good technical skills and good magerial skills in the same person. It's a blessing if you do.
This makes me worry. I am planning to upgrade my machine in a couple of months to a dual-PentiumIII. I was under the impression that the only SMP problems were at the kernel level (in development drivers and the like). But now that I hear normal apps might not be SMP safe, it makes me wonder if I would waste my money buying an SMP machine. I'm interested in hearing from people running SMP desktops as to how many problems they have caused by apps or kernel things not being SMP-safe.
Moneydance is far better than Gnucash right now. It's a little slow because it's java, but it's quite full-featured. But I'm scared to completely switch to Moneydance because I'm not sure what will happen to support if the author gets sick of it. I wish Moneydance were opensource. I'd switch in a second then.
for Gnucash to get to a similar feature level as Quicken has. Quicken is currently THE ONLY application I still need to boot into Windows for, and I wish I didn't. I keep getting very tempted to switch to Gnucash or Moneydance (which is great, but not open-source). One of the main things that I still need from Gnucash before I'll switch is auto-repeating transactions. Most of the bills I pay are the same each month, so it saves me lots of time if I just click a button, instead of entering the same data over and over again.
Yes, but Go is far more difficult for a computer to play. The best chess programs/computers in the world can play at the level of the best human chess players. On the other hand, the best computer Go programs in the world can only just barely play at the level of an average amature (about 5-10 kyu) human player. This is partly due to more attention being placed on writing good computer chess programs, but also it's due to the complexity and subtleties of Go. If you're interested, here an interesting paper on programming computer Go. http://www2.psy.uq.edu.au/~jay/go/ CS-TR-339.html For all you programmers, there is currently a prize offered by the Ing Foundation of Taiwan of about $1.5 million for any Go program that can beat a top professional human in The Ing World Computer Go Congress.
I agree. Go is far more interesting than chess. To me, chess is just one big tree search problem. Go is far more beautiful than that. There are two main online Go servers: IGS http://igs.joyjoy.net/ NNGS http://nngs.cosmic.org/ The best way to use these servers is with a client program. There are several clients available which run under Linux. My favorite is CGoban. kgo is also quite good.
It kills me every time I see a vote for school funding. It invariably gets voted down. Why? I don't understand why American citzens don't value educating their children and themselves. What better way could there be to spend tax money, than on education?
I've spent the better part of the last four months looking for a new job. I'm moving from Ireland to Oregon, so even though I have pretty good skills, it took a while to find the right place and get all of the phone interviews out of the way.
I ended up with three decent offers in the end, and one thing that I found VERY helpfull when deciding was the following question I asked all three:
"I realize that the official company policy is probably a 40 hour work week, but in reality I know that it's probably 50-70 hrs/wk. On average, how many hours do people usually work at your company?"
The internet startup said 50-60. The teleco said 45-55. The small consulting company said 40-45.
Guess which one I took? I took the consulting job, even though it offered less benefits and less stock options than the other two. I value my quality of life. If you really care about the length of your work week, then make that known to your employer or find a job where they respect that you need personal life too! In my opinion, most people are just too anxious to sacrifice their personal life for money. If you have good skills and do quality work, you can find employers that will pay you well even for 40 hour work weeks.
I recently went through compiling ghostscript with the hpdj driver for my Deskjet 660CSe and found the following in the gs-hpdj man page:
---------------------------------- FUTURE DIRECTIONS I believe I've already pushed this driver a bit beyond what is sensible. My reasons are: + First, I have already included functionality for which I do not have sufficiently reliable documentation. After several unsuccessful attempts to coax more out of HP I have given up trying to improve on this situation, and designing hardware drivers by experimentation is in my opinion simply irresponsible. One of the three main design goals of hpdj was reliability.... ---------------------------------- Lately, it appears that HP has not been as open as they used to be.
Exactly. It's the Deskjets that need better printing support. Many (most?) Laserjets already support postscript, and so don't have much of a problem printing under linux right now.
What we really need is a decent Postscript to PCL filter. Everything that's available now (ie. ghostscript) is just a reverse engineering attempt that doesn't work 100%. That, along with duplex support and cartridge alignment tools would the most usefull thing HP could do for Linux printing.
Slashdot Mirror
Not only does FreeGeek help with similar things, they were involved in this specific project too.
No, but I'll bet they can provide better Linux support than Dell.
Are you serious? A URL is an address, not necessarily something you can click on. /. left off the HREF on purpose so that people wouldn't blindly click and get burned by the malicious code.
Sure you can, if you fix the application. Complex calculations are not just a single operation, they're a series of smaller operations that are strung together resulting in a more comples argorithm. What I'm suggesting, is to alter the application so that it saves checkpoints along the way.
This reminds me of the situation that I came across while working in the computer labs in college. After people lost their paper a few times, they learned to click Save every so often. That's much easier than some fancy hybernation system people are talking about.
If you're implying that HP would be in better shape now if they had kept the T&M divisions, I disagree. Agilent is in worse share than HP is right now.
http://www.counterpane.com/crypto-gram-0111.html#
Bruce Schneier is the author of Secrets and Lies and Applied Cryptography and the inventor of Blowfish and Twofish.
There's an excellent article by Bruce Schneier in his latest Crypto-Gram newsletter discussing the issue of Full Disclosure. I recommend taking a look at it.
I've had Speakeasy as my ISP for about 18 months now, and I've been absolutely delighted with them. Not only do they not mind if I run servers off of my DSL line, but the connection has been extremely reliable. The line has been down just a few hours (one day) in the last 18 months.
There are two unix machines named A and B that are on the same subnet. Describe to me, in as much detail as possible, what happens when I type "telnet B" from a terminal on machine A.
The "in as much detail as possible" is the key phrase here. The interviewer got to see an understanding of (or lack of) PATH, inetd, DNS, subnetting, TCP/IP, ethernet, etc.
That question, and the discussion we had afterward impressed me so much about the technical caliber of the manger, I took the job.
Boy, that sounds like an interview I've had before. Have I worked with you? :)
I worked for HP for 5 years as a unix admin for manufacturing sites. I was always part of an on-call rotation. HP has what it calls GAP. It stands for Guaranteed Availability Pay. Basically, instead of paying you only when you get called in (and encouraging unreliable systems), they would pay you for being on-call. There would be no extra pay beyond GAP for a call-in.
I though this made a lot of sense because you still got paid even if there were no problems. It really encouraged you to make the systems as reliable as possible.
For that kind of money, you can afford to get a big box and an industrial strength unix like HP-UX, Solaris, or AIX. Linux is great if you don't have a lot of money and aren't going to need lots of processors and lots of RAM, but I don't think it's a good fit for what you're talking about.
It's definitely better if you can have a
technical manager, but from my experience, it's
rare. For some reason, it seems that you
never find good technical skills and good
magerial skills in the same person. It's a
blessing if you do.
This makes me worry. I am planning to upgrade my machine in a couple of months to a dual-PentiumIII. I was under the impression that the only SMP problems were at the kernel level (in development drivers and the like). But now that I hear normal apps might not be SMP safe, it makes me wonder if I would waste my money buying an SMP machine. I'm interested in hearing from people running SMP desktops as to how many problems they have caused by apps or kernel things not being SMP-safe.
Moneydance is far better than Gnucash right now. It's a little slow because it's java, but it's quite full-featured. But I'm scared to completely switch to Moneydance because I'm not sure what will happen to support if the author gets sick of it. I wish Moneydance were opensource. I'd switch in a second then.
for Gnucash to get to a similar feature level as Quicken has. Quicken is currently THE ONLY application I still need to boot into Windows for, and I wish I didn't. I keep getting very tempted to switch to Gnucash or Moneydance (which is great, but not open-source). One of the main things that I still need from Gnucash before I'll switch is auto-repeating transactions. Most of the bills I pay are the same each month, so it saves me lots of time if I just click a button, instead of entering the same data over and over again.
Yes, but Go is far more difficult for a computer to play. The best chess programs/computers in the world can play at the level of the best human chess players. On the other hand, the best computer Go programs in the world can only just barely play at the level of an average amature (about 5-10 kyu) human player. This is partly due to more attention being placed on writing good computer chess programs, but also it's due to the complexity and subtleties of Go. If you're interested, here an interesting paper on programming computer Go. http://www2.psy.uq.edu.au/~jay/go/ CS-TR-339.html
For all you programmers, there is currently a prize offered by the Ing Foundation of Taiwan of about $1.5 million for any Go program that can beat a top professional human in The Ing World Computer Go Congress.
I agree. Go is far more interesting than chess. To me, chess is just one big tree search problem. Go is far more beautiful than that.
There are two main online Go servers:
IGS http://igs.joyjoy.net/
NNGS http://nngs.cosmic.org/
The best way to use these servers is with a client program. There are several clients available which run under Linux. My favorite is CGoban. kgo is also quite good.
It kills me every time I see a vote for school funding. It invariably gets voted down. Why? I don't understand why American citzens don't value educating their children and themselves. What better way could there be to spend tax money, than on education?
Yeah. If you have to read the documentation or pay for support, they didn't make it easy enough to use, now did they?
I've spent the better part of the last four months looking for a new job. I'm moving from Ireland to Oregon, so even though I have pretty good skills, it took a while to find the right place and get all of the phone interviews out of the way.
I ended up with three decent offers in the end, and one thing that I found VERY helpfull when deciding was the following question I asked all three:
"I realize that the official company policy is probably a 40 hour work week, but in reality I know that it's probably 50-70 hrs/wk. On average, how many hours do people usually work at your company?"
The internet startup said 50-60. The teleco said 45-55. The small consulting company said 40-45.
Guess which one I took? I took the consulting job, even though it offered less benefits and less stock options than the other two. I value my quality of life. If you really care about the length of your work week, then make that known to your employer or find a job where they respect that you need personal life too! In my opinion, most people are just too anxious to sacrifice their personal life for money. If you have good skills and do quality work, you can find employers that will pay you well even for 40 hour work weeks.
I recently went through compiling ghostscript with the hpdj driver for my Deskjet 660CSe and found the following in the gs-hpdj man page:
----------------------------------
FUTURE DIRECTIONS
I believe I've already pushed this driver a bit beyond what is sensible. My reasons are:
+ First, I have already included functionality for which I do not have sufficiently reliable documentation. After several unsuccessful attempts to coax more out of HP I have given up trying to improve on this situation, and designing hardware drivers by experimentation is in my opinion simply irresponsible. One of the three main design goals of hpdj was reliability....
----------------------------------
Lately, it appears that HP has not been as open as they used to be.
Exactly. It's the Deskjets that need better printing support. Many (most?) Laserjets already support postscript, and so don't have much of a problem printing under linux right now.
What we really need is a decent Postscript to PCL filter. Everything that's available now (ie. ghostscript) is just a reverse engineering attempt that doesn't work 100%. That, along with duplex support and cartridge alignment tools would the most usefull thing HP could do for Linux printing.