Remember moisture. You don't want your cool new 10/100 8 port hub to start having problems in 6 months because there's a little condensation on the inside. If you put your stuff in a basement, think about moisture.
Conduit. Definitely conduit. I was the lead sound tech at my church for 4 years, and when we did an upgrade (we being me and one other guy), we pulled about 2 miles of microphone cable and speaker cable through tiny little holes in walls and closets. For a new house, I would suggest conduit everywhere, and bigger than you think you'll need. After all, conduit is for those unforseen wires in the future. And you will need pull-strings. A pull-string is a piece of string lain in the conduit to let you pull wire through. You tie one end of the pull-string to your new wire, and you pull on the other end to snake the wire through your conduit. Just remember to tie another sting to the back end of the wire, so that you can do it again the next week.:) I recommend 3 pull strings (fight Murphys law), and secure them to each end of the conduit; otherwise, you might drag your other strings out when you pull with one. You should seperate your power from your signal, which might mean two conduits, or leaving the power in the wall without conduit.
And in reply to the AC (Cat-5, Coaxial, etc. Cat-7 or Fibre?) who said Just don't short yourself. You won't be able to go back! - conduit lets you go back, that's its purpose. A big enough conduit, with several pull-strings, and you can run more Cat-5 when you want to turn your home office into a home-ISP.
BTW, if you plan a big setup, you might want to think about heat. A hub or two, two servers, and a patch panel in one closet might get hot enough to need cooling. I haven't run the numbers (though I am a senior in Mechanical Engineering and I know cooling), but if your closet is next to the water heater, in the middle of the summer, and your personally hosted website is mentioned on Slashdot, you might have a heat problem. Just another worry for you.:)
Good point; it seems Galileo would be a great partner if I were going to gamble in Las Vegas. However, that doesn't mean that the engineers didn't do something right without knowing it. Sometimes the most trivial, seemingly unrelated information is the crucial key to solving the problem.
I recall (but can't support) that an early physicist was conducting experiments to determine the speed of light, but his numbers were way off. He checked his instruments several times, re-checked the procedure, and still couldn't find the problem. He forgot to measure the length of the room he was conducting his experiments in; the light beam he was measuring was bouncing off of the walls before he measured it again. If he had realized what was happening, he could have accounted for the reflections more precisely.
Engineering is sometimes more art than science, and more so on the edges of the envelope. The new, untried situations are where an engineer's intuition and experience can shine brightest. But that shine rarely illuminates the situation perfectly, and improvements can always be made. We can't test our space probes in environments which correspond in any way to the environments they will experience in service, so why not test one while it is in service?
As an example of why we need to test things where we will use them, take a bar of steel and bend it. Leave that bar of steel out in an Alaskan winter. Then bend it again. It won't behave the same: it will be more brittle, fatigue weakens the crystalline structure, and you won't want to trust it ever again. ASME has an entire division devoted to Arctic Engineering, and people have lived there for centuries.
One of the hallmarks of engineering in situations where there isn't extensive experience is the acquisition and application of test data. Even where we have experience, test data is a necessity if the applicable theories are not precise or reliable enough.
Fatigue (The gradual weakening of a part [usually metal] over time. Try to break a paper-clip, you probably can't. Bend it two dozen times, it will probably break.) is a good example of an area where we have decades of experience, and we still need good test data for many situations. We need this test data. Maybe not now, but when 15 people are in the third month of a 5 year Mars mission, they will want everything to have been thoroughly tested. We can start with Galileo.
I think that the Galileo mission should be extended, both for the science, and the engineering of future spacecraft.
Galileo has outlasted the estimates of its lifespan, and I'm sure the project engineers are anxious to find out why, and more importantly, how to make the next probe more rugged. These types of probes are hard to test (zero-gravity test chamber, cosmic ray emitters, micro-meteroids, super-strong magnetic fields), so lessons about unpredicted behaviors must be learned from the previous missions. This mission has tremendous potentional to teach us about spacecraft performance in environments other than near-Earth (Hello Moon/Mars missions!), and machine performance near Jupiter (Hello colonization of Ganymede!). Oh, yeah, the science in studying Europa is important too.;) I think that NASA should continue the Galileo mission as long as the spunky little girl will keep on keeping on, if only to tell engineers how to make better spacecraft.
Signal 11's list of good proprietary software has many missing (I could add a dozen right now), but one remarkable piece of software which any programmer can learn from was not added: Mathematica, from Wolfram Research. Please bear with the off-topic paragraph, I come back, and it's even relevant.:)
In an article on GameSpy, Tim Sweeney was saying that we need a new programing language which is more intuitive and has higher level commands. His example was adding the elements of arrays A and B to make array C. Intuitively, the algorithm is just C=A+B, but in C++ it isn't nearly that simple. When I read this, I was struck (for the umpteenth time) how simple/powerful Mathematica is. In Mathematica I can just add the arrays, C=A+B, and the program takes care of the details (add element A[i,j] to element B[i,j] and assign the sum to element C[i,j]; move to element i+1,j...). This has freed me to think about the matrix math that I want to do, and I don't have to debug my matrix math algorithms when I get a screwy answer. This power and simplicity was developed under the proprietary model, and Mathematica is considered by many to be the standard advanced math software.
Relevant now:) The system of computation Mathematica uses could be a useful model to a programmer trying her hand at creating the afore mentioned 'next generation' programing language. And I had to buy it (well, I didn't _have_ to buy it, but pirating non-MicroSquish software seems unfair to the coder who wrote my new program; ethics, what a nuisance), I couldn't just download it and start solving sets of coupled partial differential equations. (How I love those PDE's.) MicroSquish/AOL isn't the only proprietary software company out there (although it seems that way sometimes), and many of those companies are staffed by people of good moral character who make money doing what they love. And they're good at it, giving us good products. (Please don't misunderstand me, I am _NOT_ saying that MicroSquish='good product', just that many proprietary software companies do a good job. Qualcomm's Eudora, for instance.)
OSS isn't the sole repository of talented programmers, there are many who work at traditional companies. Ignoring, dismissing or insulting these programmers will not produce better programs or more OSS, it will just divide the two groups even more. As the non-traditional "fringe" group, it is our responsibility to try to build a community where both sides can come to talk, and not feel discriminated against.
end rant
Louis Wu
Signal 11's comment: http://slashdot.org/comments.pl?sid=00/01/28/072 3223&cid=23
Mathematica: http://www.mathematica.com/
The GameSpy article: http://www.gamespy.com/articles/devweek_b.shtm
The Slashdot discussionof the GameSpy article: http://slashdot.org/article.pl?sid=00/01/25/1502 38&mode=thread
Slashdot Mirror
Conduit. Definitely conduit. I was the lead sound tech at my church for 4 years, and when we did an upgrade (we being me and one other guy), we pulled about 2 miles of microphone cable and speaker cable through tiny little holes in walls and closets. For a new house, I would suggest conduit everywhere, and bigger than you think you'll need. After all, conduit is for those unforseen wires in the future. And you will need pull-strings. A pull-string is a piece of string lain in the conduit to let you pull wire through. You tie one end of the pull-string to your new wire, and you pull on the other end to snake the wire through your conduit. Just remember to tie another sting to the back end of the wire, so that you can do it again the next week. :) I recommend 3 pull strings (fight Murphys law), and secure them to each end of the conduit; otherwise, you might drag your other strings out when you pull with one. You should seperate your power from your signal, which might mean two conduits, or leaving the power in the wall without conduit.
And in reply to the AC (Cat-5, Coaxial, etc. Cat-7 or Fibre?) who said Just don't short yourself. You won't be able to go back! - conduit lets you go back, that's its purpose. A big enough conduit, with several pull-strings, and you can run more Cat-5 when you want to turn your home office into a home-ISP.
BTW, if you plan a big setup, you might want to think about heat. A hub or two, two servers, and a patch panel in one closet might get hot enough to need cooling. I haven't run the numbers (though I am a senior in Mechanical Engineering and I know cooling), but if your closet is next to the water heater, in the middle of the summer, and your personally hosted website is mentioned on Slashdot, you might have a heat problem. Just another worry for you. :)
Good luck.
I recall (but can't support) that an early physicist was conducting experiments to determine the speed of light, but his numbers were way off. He checked his instruments several times, re-checked the procedure, and still couldn't find the problem. He forgot to measure the length of the room he was conducting his experiments in; the light beam he was measuring was bouncing off of the walls before he measured it again. If he had realized what was happening, he could have accounted for the reflections more precisely.
Engineering is sometimes more art than science, and more so on the edges of the envelope. The new, untried situations are where an engineer's intuition and experience can shine brightest. But that shine rarely illuminates the situation perfectly, and improvements can always be made. We can't test our space probes in environments which correspond in any way to the environments they will experience in service, so why not test one while it is in service?
As an example of why we need to test things where we will use them, take a bar of steel and bend it. Leave that bar of steel out in an Alaskan winter. Then bend it again. It won't behave the same: it will be more brittle, fatigue weakens the crystalline structure, and you won't want to trust it ever again. ASME has an entire division devoted to Arctic Engineering, and people have lived there for centuries.
One of the hallmarks of engineering in situations where there isn't extensive experience is the acquisition and application of test data. Even where we have experience, test data is a necessity if the applicable theories are not precise or reliable enough.
Fatigue (The gradual weakening of a part [usually metal] over time. Try to break a paper-clip, you probably can't. Bend it two dozen times, it will probably break.) is a good example of an area where we have decades of experience, and we still need good test data for many situations. We need this test data. Maybe not now, but when 15 people are in the third month of a 5 year Mars mission, they will want everything to have been thoroughly tested. We can start with Galileo.
I think that the Galileo mission should be extended, both for the science, and the engineering of future spacecraft.
Louis Wu
Louis Wu
Signal 11's list of good proprietary software has many missing (I could add a dozen right now), but one remarkable piece of software which any programmer can learn from was not added: Mathematica, from Wolfram Research. Please bear with the off-topic paragraph, I come back, and it's even relevant. :)
...). This has freed me to think about the matrix math that I want to do, and I don't have to debug my matrix math algorithms when I get a screwy answer. This power and simplicity was developed under the proprietary model, and Mathematica is considered by many to be the standard advanced math software.
:)
2 3223&cid=23
2 38&mode=thread
In an article on GameSpy, Tim Sweeney was saying that we need a new programing language which is more intuitive and has higher level commands. His example was adding the elements of arrays A and B to make array C. Intuitively, the algorithm is just C=A+B, but in C++ it isn't nearly that simple. When I read this, I was struck (for the umpteenth time) how simple/powerful Mathematica is. In Mathematica I can just add the arrays, C=A+B, and the program takes care of the details (add element A[i,j] to element B[i,j] and assign the sum to element C[i,j]; move to element i+1,j
Relevant now
The system of computation Mathematica uses could be a useful model to a programmer trying her hand at creating the afore mentioned 'next generation' programing language. And I had to buy it (well, I didn't _have_ to buy it, but pirating non-MicroSquish software seems unfair to the coder who wrote my new program; ethics, what a nuisance), I couldn't just download it and start solving sets of coupled partial differential equations. (How I love those PDE's.)
MicroSquish/AOL isn't the only proprietary software company out there (although it seems that way sometimes), and many of those companies are staffed by people of good moral character who make money doing what they love. And they're good at it, giving us good products. (Please don't misunderstand me, I am _NOT_ saying that MicroSquish='good product', just that many proprietary software companies do a good job. Qualcomm's Eudora, for instance.)
OSS isn't the sole repository of talented programmers, there are many who work at traditional companies. Ignoring, dismissing or insulting these programmers will not produce better programs or more OSS, it will just divide the two groups even more. As the non-traditional "fringe" group, it is our responsibility to try to build a community where both sides can come to talk, and not feel discriminated against.
end rant
Louis Wu
Signal 11's comment:
http://slashdot.org/comments.pl?sid=00/01/28/07
Mathematica:
http://www.mathematica.com/
The GameSpy article:
http://www.gamespy.com/articles/devweek_b.shtm
The Slashdot discussionof the GameSpy article:
http://slashdot.org/article.pl?sid=00/01/25/150