There's no point in packing it in a Value since it will always
be an ObjFn or ObjClosure. Leaving it an Obj* is a bit faster.
Also fixed a bug where it was freeing the closure's flexible
array of upvalues.
Added ".." and "..." infix operators. Num implements them to
return Range objects. Those in turn implement the iterator
protocol, so now numeric for loops are easy:
for (i in 1..10) { ... }
I also cleaned up the Wren benchmarks to use this.
In the process, I discovered the method_call benchmark really
just showed loop peformance, so I unrolled the loops in all of
the languages to stress method calls.
They are now invoked like "new Foo".
Also, superclass constructors are now much less semantically
and syntactically weird. Since the instance is created before
any constructor is called, there's no point in time where the
instance isn't there.
This also fixes a bug where constructors weren't being bound
correctly, and eliminates some unneeded instructions when
compiling ifs.
I also tweaked the other method_call languages to match Wren's.
Since Wren needs to do a super call to get to the parent _count,
the other languages do now too.
This is nice too because it means we're benchmarking super
calls.
It used to subtract the number of bytes during deallocation, but
that required knowing the size of an object at free time. That
isn't always available. The old code could read a freed object
while doing this. Bad!
Instead, this tracks how much memory is still being used by
marked objects. It's correct, and is also a bit less code and
faster.
Also re-ordered a few of the cases in the interpret loop.
Removing a bit of code from CODE_CLASS and CODE_CALL triggered
some branch alignment issues which had a negative perf impact.
The new ordering seems to cancel that out.
Bob Nystrom