Farray is a mess. Here's the problem:
I'm considering both an array and a pointer to it as equivalent
array values. Also pointer to an array is considered as an array object.
The idea was: you cannot modify an array value. But if your array
is in a variable you can take the address to get an array pointer.
Then like C you can still get the values by indexing, but you can
also store.
This works with get/set methods but not with index notation:
(&a) . i <- 1
because it is inconsistent. I tried this for an array or arrays:
var x : array[array[double,100],100];
(&x) . i . j <- 1
and it fails because (&x) . i is a an array of double value.
Structure address calculations propagate correctly:
struct X { x:int; };
struct Y { y:X; };
var a = Y ( X ( 1) );
println$ a . y . x;
(&a) . y . x <- 22;
println$ a . y . x;
The reason is that &a is a pointer and so &a . y is a pointer
and so &a . y . x is a pointer.
But for an array:
(&a) . i
is a value, not a pointer.
Now, after modifying the library code I got this to work:
var x : array[array[double,100],100];
for var i in 0 upto 99 do
for var j in 0 upto 99 do
var p1 : +array[double,100] = (&x).stl_begin + i;
var p2 : +double = p1.stl_begin + j;
p2 <- #rand.double / RAND_MAX . double;
done
done
In other words you can write:
((&x) . stl_begin + i) . stl_begin + j <- 1
but that\s the only way to do it (short of using compact linear types).
To make address calculations propagate,
(&x) . i
has to be a pointer, not a value. But this means &a would become
an array of pointers. It would not be writable as such so it would
be an actual array value: to get the pointed at value you'd
have to dereference.
Note this problem didn't manifest with ordinary arrays
because
p <- v
decays to
(*p) = v
i.e. even though Felix claims <- is fundamental and = is sugar
the implementation is the other way around :)
This will break some things! But it is right I think.
Actually .. it's pretty cute!!
So the change is: array[t,n] is an ArrayValue and
&array[t,n] is an ArrayValue[&t,n]
It follows:
var x = 1,2,3;
&x . 1 <- 99;
will work. note the returned value is &int so it is NOT incrementable!
--
john skaller
skal...@users.sourceforge.net
http://felix-lang.org
------------------------------------------------------------------------------
LogMeIn Rescue: Anywhere, Anytime Remote support for IT. Free Trial
Remotely access PCs and mobile devices and provide instant support
Improve your efficiency, and focus on delivering more value-add services
Discover what IT Professionals Know. Rescue delivers
http://p.sf.net/sfu/logmein_12329d2d
_______________________________________________
Felix-language mailing list
Felix-language@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/felix-language
