[OT] perl and CLRs

[Jonathan Stowe]

On Wed, 2007-05-02 at 08:49 +0100, Dirk Koopman wrote:
> Sometime ago I asked the question as to whether it was worth looking at 
> producing a CLR such as monos (or even [spit] .NET).
> 
> http://www.regdeveloper.co.uk/2007/05/01/microsoft_open_source_dot_net/
> 
> It seems to me that although the people that replied said "it couldn't 
> be done", all the other major scripting languages have (or nearly have) 
> managed it. I can't believe that producing a mono/.NET backend for perl 
> is impossible - it might be limited, it may well break modules in CPAN. 
> But it *is* the way the world is going and it does leverage work done 
> elsewhere.

I don't think I said that it couldn't be done, I just think that it
would be difficult - on a cursory examination the "Dynamic Language
Runtime" appears to be a stab at solving some of those difficulties,
however a distinct problem that Perl has that Ruby and Python don't
appear to have is that there is no strict specification of the language,
there is a single implementation which is the embodiment of Perl. The
way that I would see something like this happening is that something
closely tied to the perl implementation generated the code much like
B::Bytecode tried to do - I guess it would even be possible to shoehorn
in the IL generation from mono somehow. There is a chance that some of
the work in blead on the 5 -> 6 thing might help as well.

Of course, the flip-side to this is that the Perl community is already
working on a virtual machine and it doesn't seem that perl 5 will be
running on parrot in any complete sense any time real soon now.

> 
> Do we *really* want to continue to hold our noses in the air?

Who's holding their nose in the air? I seem to be writing more c# than
Perl in my day job these days and I also occasionally idly ponder doing
a c# on parrot thing, but there again I haven't got near to completing
the Inline::Mono thingy that I have been toying with for ages ....

/J\