OpenEuphoria

Re: Can we simplify "end"?

• Posted by Derek Parnell <ddparnell at bi?pond.co?> Aug 08, 2007
• 604 views

Bryan So wrote:
> 
> I think it is quite redundant to use different endings for
> 
>    end procedure
>    end function
>    end if
>    end while 
> 
> etc...
> 
> with proper indentation, which I am sure we all use, one
> 
>    end
> 
> should be enough to end a loop or condition or function, etc, 
> without any ambiguity.  

If the language were being designed from scratch you might have a good arguement
there. However, we have to consider all the existing code that uses the extra
token. Remember that line endings don't mean a whole lot to the interpreter. So
when we type ...

   for i = 1 to 10 do
      foo(i)
   end for

   for j = 11 to 20 do
      bar(j)
   end for

the interpreter sees ...

   for i = 1 to 10 do foo(i) end for for j = 11 to 20 do bar(j) end for

and when we remove the extra tokens ...

   for i = 1 to 10 do foo(i) end for j = 11 to 20 do bar(j) end

So if the interpreter has to support both the existing syntax and the new
syntax, whenever it sees the phrase "end for" it now has to work out Am I at the
end of a for-loop or the beginning of a for-loop? To do that it has to look ahead
a bit, back track to the right place and finalize the for-loop.

This is not impossible, but with only the current syntax or only with the new
syntax then no look-ahead/backtracking is needed.

At least two other options are available to us to get around this.

Create a new token that means 'end of loop' such as 'stop', 'done' or some such.

   for i = 1 to 10 do
      foo(i)
   done

   while sequence(X) do
      bar(X)
   done

Or create a new token that can be used regardless of the type of loop being
ended.

   for i = 1 to 10 do
      foo(i)
   end loop

   while sequence(X) do
      bar(X)
   end loop

Both of these can co-exist with the old syntax and not introduce a significant
performance hit on the intepreter.

-- 
Derek Parnell
Melbourne, Australia
Skype name: derek.j.parnell