On Mon, 2006-09-11 at 23:35 -0400, Peter Tanski wrote:
[]
Just to be sure we understand: Felix doesn't 'instantiate'
polymorphic functions during lookup.
If it did, we'd have ground types flowing through call chains,
and be passing a concrete typeclass instance through as well.
>From your comments (Peter) you seem to be expecting this,
but that's NOT how it works. C++ may do that, Felix doesn't.
When there are polymorphic functions:
fun f[t] ...
fun g[u] (x:u) => f[u] x
.. g[int] 1 ...
Then we do lookup in g for the call f[u] x.
The fact that the call g[int] will later make
u, and then t, 'int' is irrelevant: the call to
f is encoded, effectively as
f with t->u
in the body of g. The call of g is
g with u->int
The instantiator code generator actually composes
the substitutions to obtain
f with t -> u -> int
The instantiator does this to make a list of all the
instances of f, in this case just f with t-> int.
And the code generator, to replace 't' with 'int' which
in turn is replaced by the C++ name for int (also happens
to be int).
The point is, the binding of f[u] x in the body of g
is entirely independent of the fact that g[int] 1
is evaluated. The front end works like ML: all the
bindings are fully polymorphic.
--
John Skaller <skaller at users dot sf dot net>
Felix, successor to C++: http://felix.sf.net
-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
Felix-language mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/felix-language
