[nem-en] RFC: matching variant options against non-variant values

[Michal Moskal]

On 4/30/07, vc <vc at rsdn.ru> wrote:
> > I've just commited a change on trunk to disallow the following code:
> >
> > foo (o : object) : void
> > {
> >   match (o) {
> >     | [ x, y ] => ... x + y ...
> >     | _ => ...
> >   }
> > }
> > ...
>
> Michał, after this commit in Integration project appears many error
> messages.
>
> We use object fields for keep data of different types (for examle, TExpr and
> Parsetree.Fun_parm):
>
> _pexprObject = match (_pexprObject :> PExpr)
> {
> | PExpr.As    (_, Splicable.Name as name) when IsIn(name.Location)
> | PExpr.Member(_, Splicable.Name as name) when IsIn(name.Location) => ...
> | parm is P.Fun_parm when IsIn(parm.name.Location) => setobj(parm.name);
> | o => o;
> }
>
> This change duress we make many changes :(.

I don't have SVN access right now to revert it. I'll do that tomorrow.
You can comment out the check in Typer-PatternTyper.n:

+                        def parent = ti.SuperClass ().Value.GetFreshType ();
+                        when (! matched_value_type.Require (parent))
+                          ReportError (messenger,
+                                       $ "the matched value type "
+                                         "$matched_value_type was expected "
+                                         "to be subtype of $parent");

(It's important to comment out the Require() call too).

-- 
   Michał