Re: No Go To's Forever!

[Robert Dewar <robert_dewar@my-deja.com>]

In article <TnVB4.10962$624.1043257@news.flash.net>,
  "Ken Garlington" <Ken.Garlington@computer.org> wrote:
> "Robert Dewar" <robert_dewar@my-deja.com> wrote in message
> news:8b93n3$8ai$1@nnrp1.deja.com...
>
> > <<Is goto statments particular slow?>>
> > No, the concern is not for efficiency at all!
>
> You're the expert, of course, but I'm surprised by this
answer. I thought
> that structured code in general allowed for easier
determination of scope,
> making certain optimizations more likely to be applied.


Well that is true in some cases, but if gotos are simply used
to duplicate more rational control structures, then typical
optimizers are not affected. For example, for the two pieces
of code here (the original version with a goto, and the
replacement version with a loop), most optimizers will produce
identical code.

I really think that the concern with not using gotos has nothing
to do with efficiency. Indeed, it takes a very strong optimizer
to recover the efficiency of a goto in some cases. FOr example,
consider the two representations of a finite state machine

   <<state1>> .... goto state2;

   case state1 is ....  current_state := state2;

it is quite difficult for an optimizer to recover the goto
structure from the case statement, and indeed I have not worked
with any compiler that manages to eliminate the loop and case
statement in the second case.


Sent via Deja.com http://www.deja.com/
Before you buy.