From: William Rutiser <wruyahoo05@...>
Date: 2010-10-06T22:11:48+09:00
Subject: Re: how can we make a ruby compiler

Eleanor McHugh wrote:
> On 5 Oct 2010, at 22:07, William Rutiser wrote:
>   
>> As for a Ruby compiler, a useful one is very unlikely for many reasons. Now a compiler for a different language that has a distant family resemblance to Ruby may be possible but likely of not much practical use.
>>     
>
> Lisp, Scheme, Forth and Smalltalk are all compilable so in principle Ruby should be as well. It's just not clear that there's any real gain by doing so outside of very particular applications (i.e. dominated by maths or interpretation performance bottlenecks). Moreover multicore is teasing us towards a world where all processing will be a negligible cost compared to I/O latencies and throughput, much as was the case fifty years ago.
>
>   
I have just reread all the messages in this thread and decided not to 
finish a draft reply that was already too long. In summary, I argued 
that possible language implementations lie on a continuum with "pure" 
compilers and interpreters at the extremes, and most implementations for 
real languages somewhere in the middle. Therefore, it is not very 
productive to discuss whether a Ruby compiler is possible, but instead 
focus on solutions to the specific problems that cause people to ask 
about compilers for Ruby.

In this thread, only Philip Rhoades has mentioned a real application 
that he would like to be able to reprogram in Ruby and still have run 
times comparable to C. Perhaps he can expand, preferably in a new 
thread, on his program and the benefits he hopes to achieve by moving 
the work to Ruby.

-- Bill