Re: [Chicken-users] R6RS immutable pair

[John Cowan]

Kon Lovett scripsit:

> Hash-Consing?

Permitted but not required, as I say.

> You are suggesting contagion. I use this in the 'procedure-surface'  
> composite for example.

Yes.  However, there are alternative policies, like contagion towards
mutability (mutable unless all are immutable) and others.

> >Scheme would need new procedures MUTABLE-PAIR?, IMMUTABLE-PAIR?,
> >IMMUTABLE-CONS, IMMUTABLE-LIST, MUTABLE->IMMUTABLE, and
> >IMMUTABLE->MUTABLE.
> 
> Mutable vs. Immutable should be a property of the object. Sorry, but  
> I don't think we need *-immutable versions of every (<proc>  
> <datatype> ...). Just a built-in flag and exception when a mutating  
> operation attempted.

I think we're in agreement.  IMMUTABLE->MUTABLE and its opposite would
take either a list or a tree (to be decided) of one type and reconstruct
it using conses of the other type.  Otherwise, we just need the basic
constructors CONS and LIST and some new discriminators.

-- 
Income tax, if I may be pardoned for saying so,         John Cowan
is a tax on income.  --Lord Macnaghten (1901)           address@hidden