Felix can now handle all constant projections on the LHS of an assignment.
The rules are:
An *lvalue* is either
(a) a variable name (of a var)
(b) a dereferenced pointer
(c) the result of applying a primitive (C) function with "lvalue"
attribute
A valid LHS of an assignment is either
(a) an lvalue
(b) a projection applied to a valid LHS
An projection is:
(a) A field name of a struct or cstruct applied to it
(b) a record field name applied to it
(c) an integer literal applied to a tuple
(d) an integer literal applied to an array
The following are not yet supported properly but will be:
(e) An expression of a compact linear type of the type of the index of
an array applied to it
(f) an integer expression applied to an array of unitsum index
In case (f), the index expression will be bounds checked. In case (e)
of course that cannot be necessary,
In particular please note it is possible to assign into the middle
of a compact linear type, even though such a field is not addressable.
For example this currently works:
/////////////////////////////////////////
begin
println$ "Constant Double projection tuple/tuple (compact linear)";
var x = true,true,(false,case 0 of 3),true;
println$ x;
x . 2 . 1 = case 1 of 3;
println$ x;
end
begin
println$ "Constant Double projection tuple/array (compact linear)";
var x = true,true,(false,false),true;
println$ x;
x . 2 . 1 = true;
println$ x;
end
begin
println$ "Constant Double projection array/array (compact linear)";
var x = (true,true),(false,false);
println$ x;
x . 1 . 0 = true;
println$ x;
end
/////////////////////////////////////////
I know this looks trivial, and would work easily in C or whatever.
But there's a big difference. In Felix the variable "x" is all cases
is just an int.
You can think of projections of compact linear types a bit like C
bitfields. However, unlike C these can be arranged in an arbitrary
hierarchy, and, unlike C there's no restriction of such a field
to a power of 2 values. Indeed the first example contains
a trit (3 valued type).
More in next message :)
--
john skaller
skal...@users.sourceforge.net
http://felix-lang.org
------------------------------------------------------------------------------
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60134071&iu=/4140/ostg.clktrk
_______________________________________________
Felix-language mailing list
Felix-language@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/felix-language
