I run VNC over SSH over HTTP from anywhere that has a Java VM.
MindTerm has an excellent SSH client that supports proxying SSH over HTTP (CONNECT). Then I run the Java VNC client over a tunneled socket. (I even run Win200 on VMWare on top of that mess, but the cursor gets a bit confused.)
It is a bit of a dog over the wider net, but usable. On the LAN, VNC to Linux is great, but VNC to Windows isn't.
My big box ran headless for a year, with only a VNC connection.
I recommend running everything off of network drives. Setup a very stable fileserver (that you don't ever have to ship in) and run everything off of that.
It really doesn't matter what the salesman says, it really only matter how the judge interprets it. In most cases it wont get to a judge, just the threat of making a company look like it is stealing will change behavior.
Linux runs my dual 800PIII and I run W2K in VMWare to do my consulting work (work requires a W2K VPN client). Works very well for me, even cut and paste. I run emacs, ant, and the java compiler on native Linux and test in W2K on VMWare. Even when W2K is thrashing, my machine is still 50% free to do work for me.
Not only that, but I run VNC on my Linux box and via squid, open ssh, and MindTerm's excellent ssh applet, I can proxy VNC over SSH over HTTP to any workstation on the net and get to my virtual w2k box that is running on my Linux box locked in a closet. This stuff is cool.
In 12 years, I've never had to defrag a Unix FS. I defrag weekly on NT. NTFS performance goes to hell when fragmentation is bad, so you must defrag frequently, but th supplied defragger isn't schedulable.
- when I don't understand a page of a book, I go back a page and reread it. - my car has a reverse - when I drink too much beer or eat poison, I have a mechnism to get me "back". - my 16 month old daughter knows how to "put it back honey..."
I think it is hard to program for, but that's why you get paid the big bucks.
Actually, I only run Linux on my hardware. I do what works best for me. I've found that in somecases you have to pay for that, in other cases no amount of money will buy it. (in the case of OS's, you can't buy what I get with Linux (or BSD)).
Telephone Network Administration, but everyone used the acronym without cracking a smile. Of course I laughed out loud on a teleconference. I no longer work there.
Does GE move employees around to different functions?
Those firms I mentioned are all contract firms. When a customer's project ends, the firms employ may be moved into a completely different industry.
If aircraft engine sales are down, will you fluidly be transfered into light bulb manufacturing, or will you be job elimenated? I nkow that in a lot of companies, you are rather tied to a location. The company would rather job eliminate in Dallas and hire in Chicago instead of moving work and employees closer.
If you are a contractor working in GE aircraft engines and aircraft sales are down, then next month you may be placed in telecom company.
I guess my real question is, why would you ever use a closure in OO programming? Or do closures get replaced by anonymous classes? (anonymous because our implementation of Counter didn't have a name just like your lambda expression; inner is redundant)
I tried to describe the power of closures to a very OO orient IT developer and lost the battle. Every example I used was handled faster at runtime and safer in development with an OO construct.
Mainframes are slow for calculations, fast with IO. (Or more precisely very expensive for calculations and just expensive for IO.)
The IBM mainframe I used at a very large life insurance company had 11 400 Mz cpus that compared roughly to Intel CPUs, so I assume that they were a PowerPC like chip. 5 were dedicated to development and testing. One nice feature was that the L2 caches were shared across all the CPUs, unlike the Intel model, so process affinity wasn't a very big deal.
CPU resources are very expensive on a mainframe, but you do have the option of features like hot swappable CPUs and main memory. Mainframe uptimes can be 25 years; you can't do that with most computers, because you need to shut them down to upgrade CPUs and memory.
Supercomputers are another story. You can buy SGI hardware that is build around rendering images and pumping them through some very high bandwidth channels for display walls and the like. But I don't know anything about supercomputers.
If you've figured out exactly what every button and form and screen looks like, how they fit together, and how the program will work before you've written a line of code, chances arethat you are still in school.
How can you design a system with 200,000 LOC in English?! Use code for what it is good at.. describing detail. (Of course, you should use English and UML to describe design.)
The most important thing is the thing that gets delivered to the user. Normally that is an executable and sometimes user docs. What ever generates that is next important, normally code. Next important would be design.
I can't create a good executable without good code, which can't be produced without some documentation, but to say that the docs are the most important thing is dumb. I have created very good applications without any documentation. The apps were so simple that they stay in production for 5 years+.
I've done programmnig for truck engineering support.
Objects are things like engines, axles, gears.
"finite difference solutions to differential equations, finite-volume computational fluid dynamics, iterative solutions to non-linear equations, Monte-Carlo simulation of radiative heat transfer" are process to be run on objects. They may be objects also, but OO is best at helping you organize you data.
My experience with IT supporting engineers is that engineers first think of the process, then the data (data may be denormalized data sitting in some table somewhere, but here is the math that we want to perform on every row) and IT concentrates on organizing the data (here is an efficient datamodel for that data that you want to perform some operation on). See how OO would better fit the needs of the IT staff?
The trick is understanding what engineering wants to do with data, so your object models facilitate their computations, not get in the way.
The work I did was to model the truck, then using math supplied by engineering, we performed simulations with our models. The advantage was that engineering couldn't do the simulations alone because they couldn't get their minds around the vast amount of data and we couldn't do the simulations, because we didn't know the behaviors of the materials.
We acutally had objects like a transfercase that had a bracket location, gearRatios, input location, output location, currentGear, inputRPM, outputRPM, etc. Some of these values were calculated and some we given depending on the type of simulation. The trick was deciding when to calculate values and when to use cached values. (we used a dependence graph that was implicitly constructed) Are simulation would include activities like running the engine through the RPM range and each gear of the transmission, and articulate the rear axle through its positions to find binding in the drivetrain. Using these simulations we were able to select some parts (is this axle better than the previous?)
I have way too many god damn machines on a home network (windows, linux, Solaris/Sparc, OSX, etc) so simply backing up to a seperate drive doesn't look like it would cut it.
I think this would be ideal. I have a big linux box that I ba ck up to a w2k laptop, and back up the laptop and another couple desktops to the big linux box. Now a w2k virus/worm wont eat my linux box and a linux bug/worm wont eat my w2k box, so most everything is safe. My big Linux box has plenty of space to store multiple full copies of the w2k boxes.
The only problem would be if the big linux box and a desktop box were crashed at the same time. This shouldn't happen as they are at different sites (or if the explosion is that big, I don't really care about my mp3s anymore). I still have some management issues of when to delete old backups so I have multiple copies, but that's about it.
Slashdot Mirror
I run VNC over SSH over HTTP from anywhere that has a Java VM.
MindTerm has an excellent SSH client that supports proxying SSH over HTTP (CONNECT). Then I run the Java VNC client over a tunneled socket. (I even run Win200 on VMWare on top of that mess, but the cursor gets a bit confused.)
It is a bit of a dog over the wider net, but usable. On the LAN, VNC to Linux is great, but VNC to Windows isn't.
My big box ran headless for a year, with only a VNC connection.
I recommend running everything off of network drives. Setup a very stable fileserver (that you don't ever have to ship in) and run everything off of that.
Joe
Interesting. I'm reading /. waiting on our DBA's to fix our SQL server.
It really doesn't matter what the salesman says, it really only matter how the judge interprets it. In most cases it wont get to a judge, just the threat of making a company look like it is stealing will change behavior.
Joe
Sure, I do this now.
Linux runs my dual 800PIII and I run W2K in VMWare to do my consulting work (work requires a W2K VPN client). Works very well for me, even cut and paste. I run emacs, ant, and the java compiler on native Linux and test in W2K on VMWare. Even when W2K is thrashing, my machine is still 50% free to do work for me.
Not only that, but I run VNC on my Linux box and via squid, open ssh, and MindTerm's excellent ssh applet, I can proxy VNC over SSH over HTTP to any workstation on the net and get to my virtual w2k box that is running on my Linux box locked in a closet. This stuff is cool.
Joe
In 12 years, I've never had to defrag a Unix FS. I defrag weekly on NT. NTFS performance goes to hell when fragmentation is bad, so you must defrag frequently, but th supplied defragger isn't schedulable.
Maybe I just missed the ':-)'.
Joe
Why isn't back a great thing? It is very natural.
- when I don't understand a page of a book, I go back a page and reread it.
- my car has a reverse
- when I drink too much beer or eat poison, I have a mechnism to get me "back".
- my 16 month old daughter knows how to "put it back honey..."
I think it is hard to program for, but that's why you get paid the big bucks.
Joe
Actually, I only run Linux on my hardware. I do what works best for me. I've found that in somecases you have to pay for that, in other cases no amount of money will buy it. (in the case of OS's, you can't buy what I get with Linux (or BSD)).
You, my friend, are an idiot.
There is a huge difference between a residential line and a business line... quality of service. You get what you pay for.
Joe
Telephone Network Administration, but everyone used the acronym without cracking a smile. Of course I laughed out loud on a teleconference. I no longer work there.
Joe
OK. I tried Cocoon2. The design is beautifull. I hope they finish it someday, but until then it is too buggy and slow for the real world.
I was trying to build 300 page PDF reports...FOP crapped out.
I was trying to build simple screens to a DB... too many bugs in the caching layer...
For critical applications today, roll your own.
Joe
I use rsync between two large filesystems across a 4 block 802.11b link.
There is still a possibility of an intruder deleting both, but highly unlikely because one of the backups is a laptop that doesn't share resources.
[I still need a versioning filesystem, like VMS though.]
If both building are consumed in an explosion, then data recovery will be the least of my worries.
Joe
And anyone with physical access can boot off a floppy and by pass the whole burrito.
Which theif is going to know about the '-s' arguement to init?
Pretty much, if you are running Linux, expect your machine to be rebuilt with Windows.
These tracking programs are only useful for Windows, Mac, or the bios.
I'm 30, have a 14 month old and another due in May.
I am even more adament about managing my own risk now than before the kids.
Why should I trust a stranger (CEO) to take care of me and mine?
Joe
Does GE move employees around to different functions?
Those firms I mentioned are all contract firms. When a customer's project ends, the firms employ may be moved into a completely different industry.
If aircraft engine sales are down, will you fluidly be transfered into light bulb manufacturing, or will you be job elimenated? I nkow that in a lot of companies, you are rather tied to a location. The company would rather job eliminate in Dallas and hire in Chicago instead of moving work and employees closer.
If you are a contractor working in GE aircraft engines and aircraft sales are down, then next month you may be placed in telecom company.
Joe
So work for a US "keiretsus", like IBM, CTG, EDS, SIAC, etc.
Similarly, unions are essentially large contract firms added a bit of stability to the lives of the worker in trade for a slice of the pie.
Personally, I'd rather take the risk and for reward for job instability. I manage the risk personally, instead of trusting someone else.
Joe
I guess my real question is, why would you ever use a closure in OO programming? Or do closures get replaced by anonymous classes? (anonymous because our implementation of Counter didn't have a name just like your lambda expression; inner is redundant)
I tried to describe the power of closures to a very OO orient IT developer and lost the battle. Every example I used was handled faster at runtime and safer in development with an OO construct.
Joe
Do I call these Anonymous Classes and interfaces in Java?
To use your example...
interface Counter {
int getValue();
}
class MyClass {
Counter newCounter() {
return new Counter() {
int i = 3;
int getValue() { return i++; }
}
}
}
Very often the Interface is defined by the library and the anonymous implementation of the interface is written by the application programmer.
The above is type checked at compile time, so it is more difficult to abuse.
Am I missing something? This seems more obvious and less error prone than the lisp way.
Joe
Mainframes are slow for calculations, fast with IO. (Or more precisely very expensive for calculations and just expensive for IO.)
The IBM mainframe I used at a very large life insurance company had 11 400 Mz cpus that compared roughly to Intel CPUs, so I assume that they were a PowerPC like chip. 5 were dedicated to development and testing. One nice feature was that the L2 caches were shared across all the CPUs, unlike the Intel model, so process affinity wasn't a very big deal.
CPU resources are very expensive on a mainframe, but you do have the option of features like hot swappable CPUs and main memory. Mainframe uptimes can be 25 years; you can't do that with most computers, because you need to shut them down to upgrade CPUs and memory.
Supercomputers are another story. You can buy SGI hardware that is build around rendering images and pumping them through some very high bandwidth channels for display walls and the like. But I don't know anything about supercomputers.
Joe
How can you design a system with 200,000 LOC in English?! Use code for what it is good at.. describing detail. (Of course, you should use English and UML to describe design.)
The most important thing is the thing that gets delivered to the user. Normally that is an executable and sometimes user docs. What ever generates that is next important, normally code. Next important would be design.
I can't create a good executable without good code, which can't be produced without some documentation, but to say that the docs are the most important thing is dumb. I have created very good applications without any documentation. The apps were so simple that they stay in production for 5 years+.
Joe
What the hell do you do that requires opening your IDE more than once a day? Emacs starts when I log in; I don't start it again.
I've done programmnig for truck engineering support.
Objects are things like engines, axles, gears.
"finite difference solutions to differential equations, finite-volume computational fluid dynamics, iterative solutions to non-linear equations, Monte-Carlo simulation of radiative heat transfer" are process to be run on objects. They may be objects also, but OO is best at helping you organize you data.
My experience with IT supporting engineers is that engineers first think of the process, then the data (data may be denormalized data sitting in some table somewhere, but here is the math that we want to perform on every row) and IT concentrates on organizing the data (here is an efficient datamodel for that data that you want to perform some operation on). See how OO would better fit the needs of the IT staff?
The trick is understanding what engineering wants to do with data, so your object models facilitate their computations, not get in the way.
The work I did was to model the truck, then using math supplied by engineering, we performed simulations with our models. The advantage was that engineering couldn't do the simulations alone because they couldn't get their minds around the vast amount of data and we couldn't do the simulations, because we didn't know the behaviors of the materials.
We acutally had objects like a transfercase that had a bracket location, gearRatios, input location, output location, currentGear, inputRPM, outputRPM, etc. Some of these values were calculated and some we given depending on the type of simulation. The trick was deciding when to calculate values and when to use cached values. (we used a dependence graph that was implicitly constructed) Are simulation would include activities like running the engine through the RPM range and each gear of the transmission, and articulate the rear axle through its positions to find binding in the drivetrain. Using these simulations we were able to select some parts (is this axle better than the previous?)
Joe
I think this would be ideal. I have a big linux box that I ba ck up to a w2k laptop, and back up the laptop and another couple desktops to the big linux box. Now a w2k virus/worm wont eat my linux box and a linux bug/worm wont eat my w2k box, so most everything is safe. My big Linux box has plenty of space to store multiple full copies of the w2k boxes.
The only problem would be if the big linux box and a desktop box were crashed at the same time. This shouldn't happen as they are at different sites (or if the explosion is that big, I don't really care about my mp3s anymore). I still have some management issues of when to delete old backups so I have multiple copies, but that's about it.
Joe
When you bought cable service you bought a the right to use the cable network for personal use. You did not buy bandwidth.
When some punk sends a death threat to the president through your uplink, what are you going to do?
No, anonimous networks are not a utopia. Read some non fiction every once and a while.
Joe
Joe