On Fri, May 16, 2008 at 11:26 AM, Gabriel Dos Reis <[EMAIL PROTECTED]> wrote: > "Bill Page" <[EMAIL PROTECTED]> writes: > > | On Fri, May 16, 2008 at 9:10 AM, Gabriel Dos Reis wrote: > | > > | > Well, my understanding of your proposal was that in > | > > | > if cond then > | > x := 1 > | > else > | > x := 2.3 > | > ... > | > the compiler would assign type Integer to `x' in the `then' branch, > | > and type Float in the `else' branch, and type Union(Integer,Float) at > | > the meet point. > | > | No. I think *you* suggested that as an option in an earlier email. > | That was not my idea. > > My other suggestion (which I repeatedly said I was not proposing as > actual semantics, but only to point that the design space is large) > was a `real' overload of variables. Not unions. >
In http://sourceforge.net/mailarchive/message.php?msg_id=873aokc3f8.fsf%40gauss.cs.tamu.edu you wrote: "In the current semantics, branches of if-statements do not have their own scopes, so the meet environment of an if-statement is the union of environment of both branches. And this is where the error comes in: One cannot declared the same variable in the same environment with different modes." By this I understood you to be talking literally about the type 'Union', although now with further explanation I understand that you were probably thinking about the compiler "environment" in a somewhat different way. But from your comment this solution seems obvious. Regards, Bill Page. ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ open-axiom-devel mailing list open-axiom-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/open-axiom-devel
