[Date Prev][Date Next][Thread Prev][Thread Next]
[Date Index]
[Thread Index]
- Subject: Re: State machines and goto (Was: Next Version of Lua?)
- From: Alen Ladavac <alenl-ml@...>
- Date: Thu, 11 Jun 2009 14:42:23 +0200
John wrote at 6/11/2009:
> Not sure if any existing language does this, but it strikes me that
> composite loop structures would be a more elegant way of implementing this
> kind of thing:
> local done = false
> for a in foo until done do
> for b in bar until done do
> if something then done = true end
> end
> end
Two deficiencies I can see:
#1 This looks semantically the same for this case, but opens
possibility to write some hard to read code. If you write something
after that "if", and you don't use a "break" there, it will run the
code once, and then exit. How can you tell whether that was intented
or a typo (forgotten "break")?
#2 Due to ability to do #1, the compiled code would be slower, as it
has to check an additional condition in each iteration of each of the
two loops. AFAIK, as Lua byte code _does_ have a forward jump
internally, doing a real jump (either written as a goto, or as a named
break, would be more performant, I believe.
> labelled jumps are ugly and error-prone, particularly if they are limited by
> non-obvious restrictions.
Any restriction enforced by the compiler (and this one is easily
enforceable, I believe) cannot be error-prone, I think. The resulting
code is very readable, and you cannot make mistake when writing code,
as it won't compile then.
Cheers,
Alen