Re: byte-code optimizations

[Paul Pogonyshev]

> On Tue, Sep 21, 2004 at 05:05:30PM -0400, Stefan Monnier wrote:
> > What defsubst* does is treat the argument as a kind of "lexically scoped"
> > variable, but only in very limited ways.  I.e. the
>
> ...
>
> >      (defsubst* foo (x) (symbol-value x))
> >
> >      (foo y) => (symbol-value y)
> >
> > whereas our optimization won't be able to do that because it can't assume
> > a "somewhat lexically scoped" semantics.
>
> I vote for saying "you're not allowed to treat defsubst argument bindings
> as normal dynamic bindings, and if you have tons of code that does, well
> screw you, you're probably a crappy programmer anyway."

While I do agree with this, it's generally better to avoid changing
stabilized behaviour only for optimization reasons.  Otherwise,
really-difficult-to-debug bugs can jump out of nowhere.

Paul