bug, strange behaviour of libraryStack.

[Richard Gaskin]

J. Landman Gay wrote:
> Andre Garzia wrote:
>> Dar,
>> 
>> I thought of that too, no libraryStack touches the global. The first 
>> thing I thought (after hitting my head in the wall a little) was: "hey, 
>> I must be emptying the global somewhere in the libraryStack of the 
>> mainStack", but no, I never touch it...
>> 
>> its very strange. I am now double checking my code. Also, removing the 
>> 'library' call made everything work again...
> 
> The "library" command is relatively new, compared to other options. Does 
> the same thing happen if you instead use "start using stack <whatever>" 
> or "insert script of stack <whatever> into back"? These two commands 
> produce the same results and have been in the language longer. If those 
> work and "library" doesn't, then you may have found a new...feature.

There's an interesting aspect of the libraryStack message which may be a 
red herring here or may be worth exploring:

The libraryStack message is sent to a stack whenever the stack is 
brought into the message path with the "start using" command*.  But 
because the stack goes into the message path, any other stacks in the 
message path will also get the libraryStack message when it's sent to 
any stacks brought into use if that stack passes the message.

While I generally believe it's good form to pass system messages unless 
you know up front you'll absolutely never need the message downstream, 
Rob Cozens has turned me around on this one.  The libraryStack message 
is just too easy to catch more than once, and to accomodate the 
possibility that other stacks may pass it you'll want something like 
this in your libraryStack handler:

   on libraryStack
     if the short name of the target is the short name of me then
        -- do initialization sturf
     else pass libraryStack
   end libraryStack

That way a libraryStack message sent directly to it gets handled, but 
any such messages merely passed to it get passed without being handled.

Given all this, there's the possibility that another stack's 
libraryStack message is being handled in your stack, reinitializing the 
global you're having trouble with.

You could double-check that with a breakpoint, but it might be even 
easier to just add a line with an answer command to notify you when the 
message is handled:

   on libraryStack
     global gMyTroubledGlobal
     answer the short name of the target &cr& gMyTroubledGlobal
     if the short name of the target is the short name of me then
        -- do initialization sturf
     else pass libraryStack
   end libraryStack

Like I said this may be a red herring, but I've had cases of redundant 
initialization myself (before I adopted Rob's advice <g>), so it may be 
worth checking out.


* For those of you getting started with Rev and going to RevCon West, 
Dan has me scheduled to cover the message path in a session there where 
we'll discuss effective use of libraries, frontScripts, and backScripts too.

-- 
  Richard Gaskin
  Fourth World Media Corporation
  ___________________________________________________________
  Ambassador at FourthWorld.com       http://www.FourthWorld.com