Re: [moca] Ambients and functions

[Martin Berger <martinb@xxxxxxxxxxxxxx>]

Pascal Zimmer wrote:

> You might be interested by my own work: a few years ago, I designed an 
> encoding of the pi-calculus into pure safe ambients [...] However, keep 
> in mind that in both cases, the result will not be very efficient... 

I wonder ...

the decomposition of functional behaviour into name-passing seems
very efficient.  by that i mean that the translation does, or could
be made to do 'essentially the same things as conventional translations
into assembly language. i take the latter -- naively, i'm sure -- as
the relevant yard-stick for efficiency measurements. of course compilers
for functional languages employ all manner of optimisations, but they
dont seem to rely on assembly language as translation target and could
hence be coded up into name-passing as well. now what about the
efficiency of turning name-passing into named-box-movement?  it is not
clear to me how to gauge this. what do you compare against? if you have
pi-calculus or lambda calculus inside your ambients as primitive
operations (rather than as syntactic sugar) you will probably also have
to provide an elaborate infrastructure to make it work. is it so obvious
that one can do substantially better than your translation?

martin



 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The "models for mobility" mailing list     mailto:moca@xxxxxxxxxxxxxxx
http://www-sop.inria.fr/mimosa/personnel/Davide.Sangiorgi/moca.html