April 19, 2007 at 16:31LLVM 2.0 is expected by the end of the month
Why write here about a project that today barely tears the future to pieces with its irrepressible development? The fact is that you can certainly drive into the future also slowly, but seriously and for a long time.
Low Level Virtual Machine ( LLVM ) is the foundation for language developers. Maybe back-end, maybe middleware - decide for yourself.
So, yesterday in your conversations over a glass of tea with friends of IT-people, you explained that you did not develop your own advanced Lisp, Python, PHP, Perl, Ruby primarily because creating alone quickly working language implementation will not work. It was? Today you can be thrown with rotten virtual tomatoes only because of such an excuse. gcc4 transplanted to LLVM engine in most testsIt’s almost no longer behind the original gcc, and in many tests it already beats it and it beats very seriously.
But is a twofold increase in the speed of a test due to LLVM - is that a trifle worth buying? Probably not. The point is a much more far-reaching reserve. LLVM already allows standardized and high-quality mixing of languages in an application. And this is without loss of performance. And this is also right at run-time. The compiler and optimizer are always with you.
What is there to add? It is likely that LLVM is not married to a single language as a JVM, and so far 1) is not in service to a single proprietary context like the CLI from .NET. What else? Probably a few more, um, little things:
1. Outrageously neat architecture;
2. Incredibly constructive, talented and efficient project leader Chris Lattner;
3. A community similar to its leader, where there is no place for unnecessary aplomb and flames.
Take a look at them. It's worth it.
1) I want to hope that Apple still will not get this child.
Low Level Virtual Machine ( LLVM ) is the foundation for language developers. Maybe back-end, maybe middleware - decide for yourself.
So, yesterday in your conversations over a glass of tea with friends of IT-people, you explained that you did not develop your own advanced Lisp, Python, PHP, Perl, Ruby primarily because creating alone quickly working language implementation will not work. It was? Today you can be thrown with rotten virtual tomatoes only because of such an excuse. gcc4 transplanted to LLVM engine in most testsIt’s almost no longer behind the original gcc, and in many tests it already beats it and it beats very seriously.
But is a twofold increase in the speed of a test due to LLVM - is that a trifle worth buying? Probably not. The point is a much more far-reaching reserve. LLVM already allows standardized and high-quality mixing of languages in an application. And this is without loss of performance. And this is also right at run-time. The compiler and optimizer are always with you.
What is there to add? It is likely that LLVM is not married to a single language as a JVM, and so far 1) is not in service to a single proprietary context like the CLI from .NET. What else? Probably a few more, um, little things:
1. Outrageously neat architecture;
2. Incredibly constructive, talented and efficient project leader Chris Lattner;
3. A community similar to its leader, where there is no place for unnecessary aplomb and flames.
Take a look at them. It's worth it.
1) I want to hope that Apple still will not get this child.