Re: continue/repeat...until false dichotomy

["Kelley, Brian" <bkelley@...>]

On Feb 16, 2010, at 3:40 PM, Matthew Wild wrote:

> On 16 February 2010 22:24, Kelley, Brian <bkelley@qualcomm.com> wrote:
>> 
>> On Feb 16, 2010, at 1:44 PM, Florian Weimer wrote:
>> 
>> Make it:
>> 
>>  repeat
>>    local t = 0
>>    if cond then continue end
>>    local t = 1
>>    ...
>>  until t == 0
>> 
>> Then the two t are actually different variables (in the current
>> implementation), and it's not clear to which incarnation the
>> comparison refers.
>> 
>> Why is it not clear?  Would it be clearer without the continue statement?
>> The continue statement does not impact the *scope* of variables.  Scoping is
>> lexical -- as written.  Continue is just a restricted 'goto'.
> 
> If it's so clear, what is the behaviour of the code above?

Assuming "variables default to nil", when 'cond' is true and the continue is taken, "t==0" will evaluate to false because 't' will be nil.

I'm not sure what is throwing you off.  Each "local t" declares a new variable.  Here is an illustration of that:

local t
local function get() return t end
local t
local function set(a) t = a end
set(2)
print(get())  -->  nil

Inserting another "local t" above the continue statement increases the level of confusion -- as shadowing declarations usually do -- but it doesn't have any bearing on the proposed semantics of undefined variables or on how continue interacts with repeat...until.

-bhk