Re: ajax, rich text areas and frames

D. Woodill wrote:
> --- Robin Becker <robin-EZnAFvyyu2tF6kxbq+BtvQ@xxxxxxxxxxxxxxxx> wrote:
>
> > It is alo fairly easy to patch the eventloop to cause 
> > waits to timeout.
> >
> > Even if that happens there's no mechanism for waits to 
> > 'fail' (a bigger patch is required).
> >
> > --
> > Robin Becker
>
>
> How about a loop with an escape button that would
> allow Selenium users to get on with it when they
> run out of patience, like the panic button on
> midi software that you hit when all the notes stick
> on and the ugly chord makes your head want to 
> explode.   
....

certainly the patch that does the timeout can be triggered by any suitable 
event. For an automatic test the button pusher isn't available. The problem of 
causing  the wait to 'fail' is still there anyhow.

I added this to TestLoop

    this.setPollTimeout =function(t){
        if(t){
            this.pollTimeout = window.setTimeout("testLoop.checkPollTimeout()", 
t);
             }
        }
    this.checkPollTimeout=function(){
        if(this.waitForCondition){
            this.waitForCondition = function(){return true;}
            }
        }


and then changed pollUntilConditionIsTrue to look like
    /**
    * Busy wait for waitForCondition() to become true, and then continue 
command execution.
    */
    this.pollUntilConditionIsTrue = function () {
        if (this.waitForCondition()) {
            if(this.pollTimeout){
                window.clearTimeout(this.pollTimeout);
                this.pollTimeout = null;
                }
            this.waitForCondition = null;
            this.continueCommandExecutionWithDelay();
        } else {
            window.setTimeout("testLoop.pollUntilConditionIsTrue()", 10);
        }
    };


so the timing out is fairly easy with an explicit setTimeOut call on the 
testloop. It's fairly easy to generalize things to allow for a default timeout 
value for all waits. Also it seems harder to get the fact that the timeout 
occurred into pollUntilConditionIsTrue; I suppose we could set a value onto the 
testloop inside checkPollTimeout and return it if available in 
pollUntilConditionIsTrue. Assuming that we can get that information into the 
return value of pollUntilConditionIsTrue what should be done with it to indicate 
the wait failed?
--
Robin Becker