From: Elliot Temple <curi@...>
Date: 2006-07-05T12:02:38+09:00
Subject: Re: ed borasky and his second favourite programming language


On Jul 4, 2006, at 9:47 AM, Matthew Smillie wrote:

> On Jul 4, 2006, at 17:03, Nuralanur@aol.com wrote:
>
>> Ruby is "more general" than R (since there are bindings for R   
>> available),
>> and there are also bindings to other scientific  programs/packages  
>> e.g., GSL
>> (if you want to to do numerical integration or random number   
>> generation).
>
> But what does this statement mean?  Simply being "more  
> general" (which I'll just accept for now, even though the  
> definition seems a little wishy-washy) doesn't get you anywhere.   
> Being better at particular tasks is what wins.

For small, well-defined problems that won't change, a specialised  
language is best, if there is an appropriate one.

For large, well-defined specialised problems that won't change, a  
specialised language is best.

For most projects, we don't know if the specs will change, if we even  
have detailed specs in the first place. In this case when it's  
important to be able to deal with unpredictable challenges, a more  
general language is important.

R can get us to one place. Ruby can get us anywhere, a bit slower(?).  
If we're not positive where we're going, then Ruby is a good idea.

I have nothing against R, but it's not true that being better at  
particular tasks always, or even usually, wins. Even if Ruby is not  
the best at *any* particular tasks it could still frequently be the  
best choice for projects.

-- Elliot Temple
http://www.curi.us/blog/