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

[Javier Guerra <javier@...>]

On Tuesday 04 October 2005 12:10 pm, Rici Lake wrote:
> What is the semantics of 'continue'? The whole point of using
> "structured" exits like 'break' and 'continue' instead of 'goto' is
> that it syntactically prevents the error of jumping around
> initialization of a variable. So 'continue' must mean to go to the end
> of the scope, at which point the loop will be repeated. If the 'until'
> clause is inside the scope, 'continue' should continue to after the
> 'until' clause. That would be more apparent if the loop were written:
>
> repeat
>    local l = io.read"*l" or "quit"
>    if l:match"^#" then continue end
>    local g, rest = l:match"(%S+)%s*(.*)"
>    local done = handle[g](g, words(rest))
>    until done
> end
>
> which was obviously the intention. (In fact, in this particular case
> you could eliminate the local altogether.) That's what I suggested
> earlier in the thread, and the more I look at it, the more sense it
> makes to me.

i think you got 'continue' wrong.  i'd expect it to goto the loop test, not 
after it.

in fact, i feel that signaling a compiler error for this case is the best 
approach; precisely because the 'continue' means to do the check immediately. 
if a test variable isn't defined yet, it can't be right.

maybe the "can't cross scopes" check would be appropriate, but it might be 
needlessly restrictive.  with the current example:

this shouldn't compile, because the 'continue' skips over local declarations

repeat
   local l = io.read"*l" or "quit"
   if l:match"^#" then continue end
   local g, rest = l:match"(%S+)%s*(.*)"
   local done = handle[g](g, words(rest))
until done

then, the author realizes it can't be right because of the undefined test, and 
changes to this:

repeat
   local done = false
   local l = io.read"*l" or "quit"
   if l:match"^#" then continue end
   local g, rest = l:match"(%S+)%s*(.*)"
   done = handle[g](g, words(rest))
until done

now it's semantically right, but still wouldn't compile because of the "local 
g, rest"!  the only fix is to write:

repeat
   local g, rest
   local done = false
   local l = io.read"*l" or "quit"
   if l:match"^#" then continue end
   g, rest = l:match"(%S+)%s*(.*)"
   done = handle[g](g, words(rest))
until done

which is just uselessly ugly

maybe if we could think of 'continue' as syntactic sugar for a if...end block, 
any local defined after it would be in a smaller scope, that doesn't include 
the test condition.  that would explain why any local used in the condition 
have to be declared before any 'continue'; and at the same time allow for 
some 'smaller' locals after that.

<warning> i haven't ever worked the insides of any compiler, not even read the 
Lua sources; anything i say here might save us, or destroy Lua as we know it 
</warning>

-- 
Javier

Attachment: pgpjW6qbZwh1Z.pgp
Description: PGP signature