Implement method call syntax

[masak]

#25 talks about method call syntax (which is actually just a property lookup plus a function call, much like Python and JavaScript).

There's nothing preventing us from implementing those ahead of time, without having object literals. If we're going that path anyway, then we could do the property lookup/method call as a standalone thing.

Why would that matter? Because the setting currently contains a bunch of accessors/destructors for the Qtree nodes, that would be much more intuitive as methods on the Qtree nodes themselves:

value params stmts expr lhs rhs pos args ident assign block

Thus, you could do this:

my q = quasi { say("Three plus four is " ~ (3 + 4)) };
say(q.stmtlist()[0].expr().arglist()[0].rhs().lhs().value()); # 3

Instead of (as you currently have to):

say(value(lhs(rhs(arglist(expr(stmtlist(q)[0]))[0])))); # 3

This would have the further advantage of uncluttering the setting a bit more.

[masak]

Implemented in 410db3a.

[masak]

Coming back to this issue, I was surprised that I felt that all the Q properties ought to be methods. It complicates both the 007 syntax and the implementation. In fact, they're just immutable value properties. I think this was a kind of thinko, cargo-culting from subs to methods.

f244919 fixes this.

The example above now turns into

my q = quasi { say("Three plus four is " ~ (3 + 4)) };
say(q.stmtlist[0].expr.arglist[0].rhs.lhs.value); # 3

And it works under that new commit.