Re: New scoping rule for "repeat" and "continue" patch

["Mark Meijer" <meijer78@...>]

Shouldn't loops just be loops, regardless of whether the first test is done 
after the first iteration or before it? The test is actually in between 
iterations, but "repeat...until" -type loops just skip the first one (or 
alternatively "while...end" -type loops have an extra test at the start).

Repeat is an oddball because it can get in trouble with continue. Continue 
is an oddball because it can get repeat into trouble. The problem is of 
course scope, in both arguments, because continue allows you to skip a piece 
of code without enclosing that code into a new scope (which is what happens 
with the regular "if...end" that some loathe inside a loop). While I agree 
with Roberto that we don't want to turn lua into C, the link provided by 
Rici does serve to inspire:

"An iteration statement is a block whose scope is a strict subset of the 
scope of its enclosing block. The loop body is also a block whose scope is a 
strict subset of the scope of the iteration statement."

Which is exactly what Chris indicated:

>     {
>         int i;
>         for (i=0; i<10; i++)
>             ...
>     }

Wouldn't this solve our problems in lua as well? Neither repeat nor continue 
would cause any more problems, and a loop is just a loop with consistent 
scoping semantics. It would of course mean that you can't declare a variable 
local to the body of a loop and test its value in the loop condition (i.e. 
the repeat or while clause), which, it seems, is what many people want. 
Personally, I don't think this is right, but that's my 2 cents. You already 
couldn't do that with for and while loops, why should repeat be different?