Well, you never know. Maybe the front plating is an advanced form of the explosive armor they use on some tanks and armored vehicles these days; by actually exploding outward, it further reduces the force of an incoming round or warhead directed inward. Maybe going "offline" means it needs to be recharged or reloaded.
Or, maybe this really IS a deflector shield, but not yet advanced enough to be projected out, so it's sort of a surface-effect thing still. Going offline means the same thing as losing shields.
Or, maybe I need to quit doing the work of the writers and expect them to explain their own gaping plot holes.
No, it's not. 2y = y is only true for y = 0; when you divide out the y in the last step, you are dividing by 0, which is undefined, and the conclusion (2 = 1) is invalid.
Yahoo! Thanks -- I was waiting for someone to say this, and if I didn't see it, I was going to say it myself.:)
A JIT Compiler is the brute-force approach. It's almost sort of a stop-gap technique, directly speaking to the complaints that Java is slow. Like any knee-jerk reaction, it's not a really great fit.
The basic concept of a JIT Compiler is to take interpreted bytecode and compile it to native code on the off chance it might run faster. Sometimes this works, sometimes it doesn't. By the nature of what they do, server-oriented tasks get more benefit from JIT compilation than GUI apps.
Server apps tend to spend more time in one method call (e.g. the service() method of a Servlet, which sits there spitting out HTML) than do GUI apps (which, according to the event model, jump from method to method handling requests.) Since GUI apps spend very little time in one method, it doesn't gain you much to compile a particular method.
Hotspot, however, is more than a JIT compiler. It acts as a higher-level "intelligence" that monitors the execution patterns (i.e. profiles the code) and only bothers to compile the methods that would benefit from compiling. Thus, while HotSpot would spend more time reasoning out which methods need to be compiled, it will overall be more efficient, since its time will be better spent.
Crusoe's code-morphing stuff is essentially a hardware implementation of something vaguely like Java's hotspot. It serves the same purpose -- figure out what code blocks would gain the most from being compiled and then compile them.
The difference is that where HotSpot just finds "hot spots" in the code and invokes a generic compiler for the particular architecture, Crusoe has more knowledge of the VLIW core and can do better optimizations.
I'm sure there's also a lot we don't know about the details of Crusoe. There's got to be stuff they don't want the competition to know. From reading the white papers, I still have only a cursory understanding of what their code-morphing stuff is and does.
So I would say that comparing Crusoe to JIT compilation is like comparing a Ferrari to my Malibu. Comparing Crusoe to HotSpot is probably more like comparing a Ferrari to a Camaro.
I work for GE's R&D Center, where management has decided to shut down the entire facility; not just computer hardware, but everything -- even the chemistry, physics, and other experiments. The employees are physically barred from entering the facility until Monday.
This struck me personally as a tad draconian, but I can't really fault management; there's no reason to keep the facility open, especially since most of us weren't going to be there anyway.:)
They do understand the repercussions: a full shutdown means insane amounts of work just for our IMS department, let alone the actual R&D labs.
I suppose this is somewhat off-topic, since I'm talking about more than just web sites, but I thought I'd mention it and find out if any other companies are doing something similar.
You have occasionally made changes to your development efforts to accomodate non-Intel platforms; for example, making sure the latest Quake code is 64-bit clean so that it will run correctly on Alphas.
Non-Wintel markets are clearly not going to produce as much revenue as Wintel, so some people would say you're wasting time and money. What is your answer to this, and what importance do you see in multi-OS and multi-architecture development?
This is an interesting question, partly because I know a guy (a layperson) who was actually semi-clued. He's a banker, and a friend of my parents.
Back in college, I was talking to him and he asked me if I was a hacker. At the time, I was still in my professional infancy and I associated "hacker" with criminal activity, and said as much.
So, he said, "Oh right. A hacker's a guy who breaks into computers. You're a hack -- just a person who's good with computers."
So, how about we adopt the term created by a layperson? Then perhaps the laity will be more likely to adopt it.:)
> IBM just couldn't out-market Microsoft, so they capitulated. > That's why I'm using Linux now.
I hear that. I had Warp 4 installed on my Intel box (next to my Alpha) for a while but I ended up blowing away that partition for more space for Win95 so I could play games.:) (...which is all I use that box for.)
*sigh* OS/2 is too cool. Too bad it's getting left behind.
...if people,/.-ers included, would know what they're talking about before they pipe up.
Yes, it's been a year since the Mozilla project began, and no, there's no Mozilla 5.0 yet. Yes, Netscape was horsing around in the early days, and released 4.5 before comitting to Mozilla seriously and wasted 6 months. Yes, there are only around 30 people outside Netscape who work on Mozilla.
So what? That's a 1/3 increase in the developer base, for free. It's still a success in terms of money, even if it hasn't taken off as a major open source project.
As for you idiots spouting off about Netscape releasing "crap", try looking at the code recently. Mozilla IS a brand-new browser; very little of the code remains from 4.x at this point. The reason Mozilla 5.0 is not yet released is because Mozilla 5.0 has only been in the works for 6 months!
The first 6 months were, granted, wasted. That was a mistake, and an acknoledged one (by jwz at least.) The days when you can write a modern respectable web browser in six months went out with Navigator 1.0, however, so give them some time. They're moving quickly with it, and for what Mozilla 5.0 offers they're moving damned fast.
I know this is probably too much to ask, but if you don't understand the software engineering process or the way businesses look at the open source model, then please just stay quiet. You can't run around calling things like Mozilla a failure until they bust.
Slashdot Mirror
Well, you never know. Maybe the front plating is an advanced form of the explosive armor they use on some tanks and armored vehicles these days; by actually exploding outward, it further reduces the force of an incoming round or warhead directed inward. Maybe going "offline" means it needs to be recharged or reloaded.
Or, maybe this really IS a deflector shield, but not yet advanced enough to be projected out, so it's sort of a surface-effect thing still. Going offline means the same thing as losing shields.
Or, maybe I need to quit doing the work of the writers and expect them to explain their own gaping plot holes.
No, it's not. 2y = y is only true for y = 0; when you divide out the y in the last step, you are dividing by 0, which is undefined, and the conclusion (2 = 1) is invalid.
Sun released Tomcat (the Servlet/JSP reference implementation) under the Apache license as Project Jakarta. This was a pretty major release. :)
Check it out at:
jakarta.apache.org.
A JIT Compiler is the brute-force approach. It's almost sort of a stop-gap technique, directly speaking to the complaints that Java is slow. Like any knee-jerk reaction, it's not a really great fit.
The basic concept of a JIT Compiler is to take interpreted bytecode and compile it to native code on the off chance it might run faster. Sometimes this works, sometimes it doesn't. By the nature of what they do, server-oriented tasks get more benefit from JIT compilation than GUI apps.
Server apps tend to spend more time in one method call (e.g. the service() method of a Servlet, which sits there spitting out HTML) than do GUI apps (which, according to the event model, jump from method to method handling requests.) Since GUI apps spend very little time in one method, it doesn't gain you much to compile a particular method.
Hotspot, however, is more than a JIT compiler. It acts as a higher-level "intelligence" that monitors the execution patterns (i.e. profiles the code) and only bothers to compile the methods that would benefit from compiling. Thus, while HotSpot would spend more time reasoning out which methods need to be compiled, it will overall be more efficient, since its time will be better spent.
Crusoe's code-morphing stuff is essentially a hardware implementation of something vaguely like Java's hotspot. It serves the same purpose -- figure out what code blocks would gain the most from being compiled and then compile them.
The difference is that where HotSpot just finds "hot spots" in the code and invokes a generic compiler for the particular architecture, Crusoe has more knowledge of the VLIW core and can do better optimizations.
I'm sure there's also a lot we don't know about the details of Crusoe. There's got to be stuff they don't want the competition to know. From reading the white papers, I still have only a cursory understanding of what their code-morphing stuff is and does.
So I would say that comparing Crusoe to JIT compilation is like comparing a Ferrari to my Malibu. Comparing Crusoe to HotSpot is probably more like comparing a Ferrari to a Camaro.
- Dan
This struck me personally as a tad draconian, but I can't really fault management; there's no reason to keep the facility open, especially since most of us weren't going to be there anyway. :)
They do understand the repercussions: a full shutdown means insane amounts of work just for our IMS department, let alone the actual R&D labs.
I suppose this is somewhat off-topic, since I'm talking about more than just web sites, but I thought I'd mention it and find out if any other companies are doing something similar.
Any other employers doing full shutdowns?
You have occasionally made changes to your development efforts to accomodate non-Intel platforms; for example, making sure the latest Quake code is 64-bit clean so that it will run correctly on Alphas.
Non-Wintel markets are clearly not going to produce as much revenue as Wintel, so some people would say you're wasting time and money. What is your answer to this, and what importance do you see in multi-OS and multi-architecture development?
Back in college, I was talking to him and he asked me if I was a hacker. At the time, I was still in my professional infancy and I associated "hacker" with criminal activity, and said as much.
So, he said, "Oh right. A hacker's a guy who breaks into computers. You're a hack -- just a person who's good with computers."
So, how about we adopt the term created by a layperson? Then perhaps the laity will be more likely to adopt it. :)
hacker == criminal
hack == guru/wizard/correct "hacker"
I guess the grammar's not right, but I thought I'd mention it. Other alternatives include guru, wizard, paragon...
Hmm... tough one.
> That's why I'm using Linux now.
I hear that. I had Warp 4 installed on my Intel box (next to my Alpha) for a while but I ended up blowing away that partition for more space for Win95 so I could play games.
*sigh* OS/2 is too cool. Too bad it's getting left behind.
Yes, it's been a year since the Mozilla project began, and no, there's no Mozilla 5.0 yet. Yes, Netscape was horsing around in the early days, and released 4.5 before comitting to Mozilla seriously and wasted 6 months. Yes, there are only around 30 people outside Netscape who work on Mozilla.
So what? That's a 1/3 increase in the developer base, for free. It's still a success in terms of money, even if it hasn't taken off as a major open source project.
As for you idiots spouting off about Netscape releasing "crap", try looking at the code recently. Mozilla IS a brand-new browser; very little of the code remains from 4.x at this point. The reason Mozilla 5.0 is not yet released is because Mozilla 5.0 has only been in the works for 6 months!
The first 6 months were, granted, wasted. That was a mistake, and an acknoledged one (by jwz at least.) The days when you can write a modern respectable web browser in six months went out with Navigator 1.0, however, so give them some time. They're moving quickly with it, and for what Mozilla 5.0 offers they're moving damned fast.
I know this is probably too much to ask, but if you don't understand the software engineering process or the way businesses look at the open source model, then please just stay quiet. You can't run around calling things like Mozilla a failure until they bust.