There is nothing wrong per se with rewriting code from the ground up: it's generally pretty quick and easy compared to the original project, and you can reuse large chunks of code. The problem is the second system effect, where you don't just rewrite, but try to create a vastly more complex and powerful second system. That's what seems to have happened with Mozilla.
A high learning curve doesn't mean that the systems are "hard" to extend or modify, it means that it's hard to learn to
extend and modify.
Well, I think it's both for Gnome, KDE, or other large C/C++ systems.
Aren't MS using some big virtual machine for their common runtime? I'm concerned about performance, mostly ) I'm also
suspicious of GC.
Well, the short answer is: your suspicions and concerns are unfounded. The long answer is that as long as people don't start using this stuff, it will continue to appear slow. The only reason why C/C++ seems "fast" is because many programmers are familiar with it and code to its performance limitations, and because operating systems preload dozens of megabytes of shared libraries.
I didn't really get what you were trying to say about GNOME. Are you trying to say that CORBA doesn't have
exceptions, or an object model or type information anything like that ? IIRC, GNOME has all the things you claimed it
didn't have (though I don't program in GNOME that much -- I prefer KDE). QT certainly has its own rtti system, its own object model, and a CORBA-like system (DCOP) which acts as a cross-language object system.
Whatever Gnome and Qt call it, those systems simply don't provide the functionality and guarantees that Microsoft's common runtime or the JVM provide. And they require a lot more work on the part of the programmer.
Again, subtle M$ bullshit propaganda. [...] Happy New Year q000921
Make it your new years resolution to leave the dark side!
Given that I've used Linux since pre-1.0 releases and GNU software since the 80's, and introduced GNU/Linux at half a dozen organizations, that's a pretty silly suggestion.
However, I don't let my long involvement with Linux blind me to when Microsoft is doing something right. If they follow through with using the common runtime as the basis for most of their software, they will have a huge advantage over Linux and its C/C++-based systems.
I don't intend to start using Microsoft because of.NET. But I expect that the open source software I will be using instead will not be Gnome or KDE either. While those are excellent short term solutions, in my opinion, they just won't remain useful in the long run, for pretty much the same reasons Microsoft is looking for a new approach now.
Mozilla 6 couldn't have been constructed any [more] differently than IE,
and Gnome has always been radically different than anything M$
has ever conceived.
Gnome, KDE, and Mozilla are large C/C++ systems, and they both try to force a dynamic object system on top of those languages. It's a testament to human persistence in the face of great odds that those systems are as nice as they are. But it's also pretty clear that those systems are neither easy to extend nor easy to modify--the learning curve is pretty high.
I suggest you make it your New Year's resolution to learn a bit more about systems other than Linux, Windows, and C/C++. In fact, in the 70's and 80's, there were a lot of good ideas and developments GUI systems, object systems, languages, and kernels, compared to which Gnome and the Linux kernel look primitive.
Slashdot Mirror
There is nothing wrong per se with rewriting code from the ground up: it's generally pretty quick and easy compared to the original project, and you can reuse large chunks of code. The problem is the second system effect, where you don't just rewrite, but try to create a vastly more complex and powerful second system. That's what seems to have happened with Mozilla.
Well, I think it's both for Gnome, KDE, or other large C/C++ systems.
Well, the short answer is: your suspicions and concerns are unfounded. The long answer is that as long as people don't start using this stuff, it will continue to appear slow. The only reason why C/C++ seems "fast" is because many programmers are familiar with it and code to its performance limitations, and because operating systems preload dozens of megabytes of shared libraries.
Whatever Gnome and Qt call it, those systems simply don't provide the functionality and guarantees that Microsoft's common runtime or the JVM provide. And they require a lot more work on the part of the programmer.
Given that I've used Linux since pre-1.0 releases and GNU software since the 80's, and introduced GNU/Linux at half a dozen organizations, that's a pretty silly suggestion.
However, I don't let my long involvement with Linux blind me to when Microsoft is doing something right. If they follow through with using the common runtime as the basis for most of their software, they will have a huge advantage over Linux and its C/C++-based systems.
I don't intend to start using Microsoft because of .NET. But I expect that the open source software I will be using instead will not be Gnome or KDE either. While those are excellent short term solutions, in my opinion, they just won't remain useful in the long run, for pretty much the same reasons Microsoft is looking for a new approach now.
Gnome, KDE, and Mozilla are large C/C++ systems, and they both try to force a dynamic object system on top of those languages. It's a testament to human persistence in the face of great odds that those systems are as nice as they are. But it's also pretty clear that those systems are neither easy to extend nor easy to modify--the learning curve is pretty high.
I suggest you make it your New Year's resolution to learn a bit more about systems other than Linux, Windows, and C/C++. In fact, in the 70's and 80's, there were a lot of good ideas and developments GUI systems, object systems, languages, and kernels, compared to which Gnome and the Linux kernel look primitive.