I agree 100%. Write code that is easy to understand and modify, then optimize it, but only after you have profiled it to find out where optimization will actually matter.
IMHO, software has gotten significantly less usable in the 15 or so years that I have been using computers on a regular basis. Several examples of this trend:
Skins. Goodbye consistency. I blame Winamp for this, although Windows Media Player has taken the problem to dizzying new heights.
Software that tries to outguess the user. E.g., menu items that aren't shown by default, because you "probably don't need to use them".
Feature bloat. Firefox is the only mainstream application I can think of that actually removed features from an application in order to improve usability. Bravo.
Linux-based digital music players are a great idea, but the manufacturers don't always honor the GPL by making the source available. For example, iRiver appears to be violating the GPL with its PMP-140 product:
Kildall wrote a seminal paper called "A Unified Approach to Global Program Optimization" which introduced dataflow analysis as a general technique for program analysis and compiler optimization. Every time you add -O([1-6])* to your gcc command line, you're applying techniques that Kildall invented.
CP/M was pretty cool, too:-)
Important differences between Java and C#
on
Java 1.5 vs C#
·
· Score: 5, Insightful
While neither Java nor C# is truly free of being controlled by an Evil Corporation(tm), Java at least has multiple vendors, runs on a wide variety of platforms, and has an open standardization process.
The idea of machine code to machine code translation is not new. For example, Dynamo appeared in 1999, although in its case the source and target instruction sets were the same. Regardless of whether or not transitive is for real (I'm guessing it is), it's clearly just a matter of time before this technology appears in a practical form.
I disagree. IBM has put a huge amount of time and resources into free software that lots of people use: Eclipse, Linux, etc. The effort to get Eclipse to work on Jikes RVM gives users a world-class Java IDE which is entirely based on open source software. One of the most interesting things about Eclipse on Jikes RVM is that it does not require any changes to Eclipse; the porting effort consistent entirely of fixing bugs and implementing missing functionality in Jikes RVM and its libraries.
Disclaimer: As an intern at IBM, I was one of the people involved in the effort to get Eclipse to run on Jikes RVM. I don't work for them at the moment.
I started out running Linux in January of 1994, using MCC on my 25MHz 486SX with 4 MB of RAM. I think the entire distribution was about 7 or 8 1.44 MB floppies, which I painstakingly downloaded using bitftp. (Yes, I was on bitnet. If you don't know what bitnet was, be very grateful).
The machine was too underpowered to run X comfortably (although I did play around a bit with TinyX - made the machine swap like nobody's business); however, I hooked up my VT102 terminal in order to have separate vi and bash sessions. Just poking around trying to figure out what all of the Unix programs did was great fun. yacc? What the hell is that? The best part was gcc; my previous programming experience had been with Microsoft Quick C. Having a development system where my buggy programs didn't cause the entire system to crash was a beautiful thing.
Anyway, those were good times, and it's nice to see that these old distributions are still around.
Microsoft's contribution will not be monetary, but services worth millions of dollars, including a full-time on-site project manager, planning and design expertise, staff training and ongoing technology support.
I think it would be great if they were donating $46M to build a school, but they're not.
This has been done before!
on
X11 in ASCII
·
· Score: 1
IIRC, the original port of X11 to Linux was done by Orest Zborowski. I seem to remember reading an interview in which he said that to simplify porting, he first did a 1-bit visual which used the 80x25 text screen.
Classpath is just the Java libraries. Although quite a bit of the standard Java APIs are implemented, there are some major missing pieces - the biggest is Swing, which would be a huge undertaking to reinvent.
My impression from reading the article is that Redhat has been developing a virtual machine. Although there are already a some fairly high quality open source VMs out there (e.g., Jikes RVM), having another one would be cool, especially with a company like Redhat behind it.
I'm glad I started grad school in '98, when you basically just had to know how to (1) speak English and (2) breathe. If I had to apply now, I doubt I'd get in!
These new java features or shortcuts whatever reminds of C++... Is Java going to come with "Pointer" manipulation features later on?
No. Although the parametized types in Java have a superficial resemblance to C++ templates, they compile down to bytecode which is more-or-less identical to what you would get if you just used non-parametized types and lots of casts (i.e., the old way of doing things.) I have used Java with generics for almost a year now, and trust me, once you've used them there's no going back. They rule.
Will java become the next C++ and will be extremely tidious to program with? Overall, change is good...:)
Java will not be extended in any way that would compromise type safety or backward compatibility. That basically precludes inheriting any real brain damage from C++.
There is a substantial amount of current research which focuses on identifying defects in software. For example, many synchronization errors, buffer overflow errors, etc. can be detected using automated techniques. Surprisingly, a significant number of bugs can be found even in well tested commercial code. My advisor and I are working on a tool for finding bugs in Java code:
We've found that even very simple bug pattern detectors turn up dozens or hundreds of bugs in production code.
The good news is that as bug-finding techniques mature and become more widespread, more bugs can be found during development rather than
after applications are deployed.
One reason I would not feel comfortable programming a large project in a scripting language is the lack of static type checking. Compile-time guarantees about runtime behavior are a very, very good thing.
Someone should invent a scripting language with static type checking. (Are there any?)
I don't know about IBM buying Sun, but regarding IBM and Java, something that might interest the Slashdot community is Jikes RVM, a high performance Java implementation written in Java. It's been the basis of some very interesting research, and best of all it's free software.
Slashdot Mirror
I do agree that the Motif version looks like ass (and is a pain to use), but then, that's Motif's fault, not SWT.
I agree 100%. Write code that is easy to understand and modify, then optimize it, but only after you have profiled it to find out where optimization will actually matter .
Kildall wrote a seminal paper called "A Unified Approach to Global Program Optimization" which introduced dataflow analysis as a general technique for program analysis and compiler optimization. Every time you add -O([1-6])* to your gcc command line, you're applying techniques that Kildall invented.
:-)
CP/M was pretty cool, too
While neither Java nor C# is truly free of being controlled by an Evil Corporation(tm), Java at least has multiple vendors, runs on a wide variety of platforms, and has an open standardization process.
Well, when I say 'house' it was only a hole in the ground covered by a sheet of tarpaulin, but it was a house to us.
The idea of machine code to machine code translation is not new. For example, Dynamo appeared in 1999, although in its case the source and target instruction sets were the same. Regardless of whether or not transitive is for real (I'm guessing it is), it's clearly just a matter of time before this technology appears in a practical form.
When I was growing up in the Syracuse area we used to call it the "Sub-Standard". Ah, good times.
Anyway, I'm sure "real" encyclopedias never contain any errors.
High taxes, inhumanly cold temperatures in winter...what more could anyone want? :-)
- Mobius
- O3one
Happily, he did mention my hobby OS.Emulators like VirtualPC and Bochs are a really nice way to play with operating system code without having to worry about screwing up your machine.
Disclaimer: As an intern at IBM, I was one of the people involved in the effort to get Eclipse to run on Jikes RVM. I don't work for them at the moment.
I started out running Linux in January of 1994, using MCC on my 25MHz 486SX with 4 MB of RAM. I think the entire distribution was about 7 or 8 1.44 MB floppies, which I painstakingly downloaded using bitftp. (Yes, I was on bitnet. If you don't know what bitnet was, be very grateful).
The machine was too underpowered to run X comfortably (although I did play around a bit with TinyX - made the machine swap like nobody's business); however, I hooked up my VT102 terminal in order to have separate vi and bash sessions. Just poking around trying to figure out what all of the Unix programs did was great fun. yacc? What the hell is that? The best part was gcc; my previous programming experience had been with Microsoft Quick C. Having a development system where my buggy programs didn't cause the entire system to crash was a beautiful thing.
Anyway, those were good times, and it's nice to see that these old distributions are still around.
IIRC, the original port of X11 to Linux was done by Orest Zborowski. I seem to remember reading an interview in which he said that to simplify porting, he first did a 1-bit visual which used the 80x25 text screen.
Blackdown is a port of Sun's JDK. Not open source at all.
Classpath is just the Java libraries. Although quite a bit of the standard Java APIs are implemented, there are some major missing pieces - the biggest is Swing, which would be a huge undertaking to reinvent.
My impression from reading the article is that Redhat has been developing a virtual machine. Although there are already a some fairly high quality open source VMs out there (e.g., Jikes RVM), having another one would be cool, especially with a company like Redhat behind it.
I'm glad I started grad school in '98, when you basically just had to know how to (1) speak English and (2) breathe. If I had to apply now, I doubt I'd get in!
No. Although the parametized types in Java have a superficial resemblance to C++ templates, they compile down to bytecode which is more-or-less identical to what you would get if you just used non-parametized types and lots of casts (i.e., the old way of doing things.)
I have used Java with generics for almost a year now, and trust me, once you've used them there's no going back. They rule.
Java will not be extended in any way that would compromise type safety or backward compatibility. That basically precludes inheriting any real brain damage from C++.
We've found that even very simple bug pattern detectors turn up dozens or hundreds of bugs in production code.
The good news is that as bug-finding techniques mature and become more widespread, more bugs can be found during development rather than after applications are deployed.
One reason I would not feel comfortable programming a large project in a scripting language is the lack of static type checking. Compile-time guarantees about runtime behavior are a very, very good thing.
Someone should invent a scripting language with static type checking. (Are there any?)
I don't know about IBM buying Sun, but regarding IBM and Java, something that might interest the Slashdot community is Jikes RVM, a high performance Java implementation written in Java. It's been the basis of some very interesting research, and best of all it's free software.
For those in the D.C. area, NASA Goddard has surplus property auctions. I've never been, but it looks like they have some pretty cool junk.