__str__ vs. __repr__

[Bjorn Pettersen]

> M.-A. Lemburg <mal at lemburg.com> wrote:
> > > You may be wondering how this works. The answer is that 
> Viperi does not
> > > use bytecode, it executes the 'Abstract Syntax Tree' directly.
> > > Consequently, the original code can be recovered. 
> (without comments,
> > > or the original formatting, of course).
> > 
> > Could you point me to some resources ? Using ASTs for execution
> > is an interesting subject and I would like to know how you deal
> > with Python dynamic nature in this context (are the ASTs self
> > modifying ?).
> 
> guess my brain doesn't work well today,
> but maybe someone could tell me:
> 
> 1) what's the reason AST's would have to change during
> the execution of dynamic code (the byte code doesn't
> exactly change, does it?)

It doesn't have to. All you have to do is implement an "Environment" object
that maps variable names to objects in the current scope. E.g.:

	def eval_assignment(lhs, rhs, env):
		left = evaluate(lhs)
		right = evaluate(rhs)
		env[left] = right

Most simple interpreters work this way (unless you've spent a lot of time on
implementing your bytecode interpreter, it is generally a very modest win).

> 2) what's the reason byte code cannot be used to recover
> the original source code (Swedish readers may remember
> the ABC80, which did exactly this).

As long as the bytecode is relatively close to the language, this is no
problem. (You can even do this for Java bytecodes..)

-- bjorn