Is this sort of thing a bug?

[Richard Gaskin]

David Bovill wrote:
> There is a case, where the long id stops working which was intoduced since
> 3.5 and behaviors. Though I have not had time to track down the cause. I
> have filed it here <http://quality.runrev.com/qacenter/show_bug.cgi?id=8014>
> 
> It totally baffles me but sometimes "the long id of the target" fails to
> convert an object reference form the form "control id 1234 of stack XXX" -
> I've set a breakpoint in a crucial handler so I can clearly see the script
> and variables when it fails - but can;t reproduce it except to note it only
> happens occasionally with this behavior type reference. Anyway...

Very interesting bug you stumbled across there.  Oddly enough, I can't 
reproduce it using:

   put the long id of btn id 1008 of stack "b"

...which returns the object descriptor with the full hierarchy as we 
would expect:

   button id 1008 of card id 1002 of stack "b" of stack 
"/Users/rg/Desktop/a.rev"

Maybe this issue is specific to debug mode?  Have you found a recipe for 
getting that result outside of debugging?


The weirder part of that report was the part about behaviors being 
disconnected even though the behavior reference was apparently correct:

    Then I issued "set the behavior of grp 1 to the behavior
    of grp 1" and everything started working again across
    multiple stacks.

If the behavior is a valid reference, what would cause it to go bad?

I agree with your suggestion there that if the engine is unable to find 
a referenced behavior it should automatically do the equivalent of "set 
the behavior to the behavior" to attempt to fix it before bailing out.


> What would be the ideal solution to make short and long IDs of stacks most
>> consistent with the rest of the language?
> 
> I'd go for returning the short name of the stackas equivalent to a unique
> "id" of an object - unless there is a use for the current id (number)
> returned.

I'm inclined to agree, but it may be useful to know why stacks have 
integer IDs at all.

--
  Richard Gaskin
  Fourth World
  Revolution training and consulting: http://www.fourthworld.com
  Webzine for Rev developers: http://www.revjournal.com