Hi Jecel,

To me it's not a question of optimisation, but a question of not being
able to express useful patterns of control flow directly. Instead, I am
forced to encode them, often using otherwise-avoidable mutable state.

Imagine a BASIC without GOSUB. You have to use arrays to represent a
stack if you need to implement a push-down automaton. It's a very
similar situation here: state machines have a natural implementation
using tail calls that is not available unless dead frames are elided.
The alternative is to use an instance variable or similar to hold the
current state, and to perform explicit trampolining.

Regards,
  Tony


Jecel Assumpcao Jr wrote:
> Tony Garnock-Jones wrote:
>> sig wrote:
>>> But then, what to deal with such cases:
>>>
>>> aClass>>add: a with: b
>>> ^ self add: a with: b+1
>> This is a call in tail position, and it's a crying shame that the Squeak
>> VM doesn't presently allow the associated degenerate stack frames to be
>> elided.
> 
> Lukas Renggli gave a series of examples where replacing the "tail send"
> with a "goto" would give use the wrong semantics. That, however, is a
> separate issue from avoiding the creation of stack frames.
> 
> Dan Ingalls said that VM implementors can cheat as long as they don't
> get caught and if you ever get into the debugger you will certainly
> notice the lack of some stack frames and will curse whoever had the
> brilhant idea of eliminating them to get some extra speed. The Self guys
> were very fanatical about not getting caught this way no matter what the
> costs in terms of performance.
> 
> For tinySelf 1, however, the cost in performance would have been
> extremely high since having tail send optimization made programs far
> more parallel than they would be otherwise. Unfortunately this single
> feature was responsible for practically all the hard to find bugs in the
> implementation and so when I had to move on to another project they
> became the reason why tinySelf 1 remained unfinished. Given this
> experience I think it would have been better for me to first have made
> the system fully functional without this optimization and then add it in
> a second version. But I would have added it.
> 
> - -Jecel
> 
>