[OT] perl and CLRs (.NET rocks)

[Dirk Koopman]

Ovid wrote:
> --- Dirk Koopman <djk at tobit.co.uk> wrote:
> 
>> However, anybody that does "scripting" in either of these two
>> languages 
>> is missing the point (as well as being a potential RSI victim). This
>> is 
>> why I am, very gently, agitating for my favourite scripting language
>> to 
>> have standard backend, preferably .NET/Mono but I could live with
>> JVM.
> 
> Jonathan Worthington, the gentleman who got .NET running on Parrot and
> thus should be in a position to know, argues that .NET and JVM are
> generally bad targets for Perl as they are not optimized for dynamic
> languages (meaning, in this case, where the data are typed rather than
> the variables) and would need a fair bit of support built into them to
> make Perl run as smoothly as it could.  How IronPython and Ruby.NET
> have managed to bridge this gap, I couldn't say, but I'd trust
> Jonathan's judgment as much as anyone's.
> 

This was the comment I got when I mentioned it previously, however it 
should not be forgotten that the average CPU is likely to be even less 
"optimised".

I imagine the real reason is that perl has a much more incestuous 
relationship with its interpreter than most other languages. The fact 
that the basic interpreter loop traverses a linked list, which can be 
altered - at run time - at will, does not help us any and I imagine that 
is why we have yet to see *any* non-perl backend that works 100%. And 
this seems to include parrot et al.

Just because it "isn't optimised" does not, by itself, seem like a good 
reason for not trying. And at least we have a fairly well defined, 
documented and static target to hit.

Dirk