Re: [sigc] Handling signal loops

["Andrew E. Makeev" <andrew solvo ru>]

Timothy M. Shead wrote:

> On Mon, 2005-09-26 at 10:25 +0400, Andrew E. Makeev wrote:
>  
>
> > Timothy M. Shead wrote:
> >
> >    
> >
> > > I'm wondering if anyone has come across a way to handle signal "loops" -
> > > that is to say, if signal A is connected to signal B, and B to A, is
> > > there any way to prevent emission of either from entering an infinite
> > > loop?  I tried creating a custom accumulator, but discovered that a new
> > > instance of the accumulator is created every time emit() is called.  I
> > > know it sounds crazy, but I'd like to find a way to see that every
> > > signal in a given loop gets called, then the loop is broken.
> > >
> > >
> > >
> > >
> > >      
> > Try static member for accumulator class, and increase it in each signal 
> > callback, so, when you hit "odd" number, for example, stop the signal 
> > emission with accumulator return state.
> > Guess, that's clear.
> >    
>
> Seems like I could use this to impose an arbitrary limit on the number
> of emissions that take place, and thus prevent infinite loops, but I
> would have know way of knowing whether I'd hit every signal
> once-and-only-once, or whether I terminated too soon ... unless I'm
> missing something?
>  
Using static flags you could do whatever you would to.
One way:
flag = 0 - no signal
flag = 1 - 1st signal
flag = 2 - 2nd signal
When enter 1st slot, check if (flag == 0 || flag == 2) else (duplicate, 
stop). if (flag == 2) then (flag = 0, stop) else (flag = 1).
When enter 2nd slot, check if (flag == 0 || flag == 1) else (duplicate, 
stop). if (flag == 1) then (flag = 0, stop) else (flag = 2).