Could you include the error message that you get when Segue gives up the Ghost? Also do you have a window title change or another context such that the title or tag identifier for the window may change.
because my guess is the way you are recognizing the main window either by:
(2) x and y coordinates (which you should never do)
(3) window id
Whatever is changing in the middle of your test and your old usefull frame file .. is not keeping up with the change. You want to really focus in on any and all changes that take place in the app from point "good show" to poing "say whaaaat!"
My suggestion as soon as the script fails .. re-record that window and compare the tags starting at the highest level. You could also use the Record >> Window Identifiers .. and see where the taggingh breaks. In other words from the script you have run that and start placing the cursor over various window elements .. when everything is good the identifier "should" map to your frame file. When it breaks you will see the identifier get generated by the system.
However make sure you save your script and hit the compile (checkmark) logo. Even a minor change in your test script will require you to do this .. this will ensure that the second method outlined will work consistantly.
Finally if this is all greek to you send us the frame file definition of ONLY the ONE window that fails. Give us teh one that worked all along, then rerecord that one window and send it over as well. If the problem I described is occuring, any number of helpfull people in these forums will spot it preety quickly.
This is a question for the experts. I really need to know the process of window object recognition, because one of our products I am testing [img]images/icons/rolleyes.gif[/img] seems to be "blocking" that recognition - meaning - at one point of time, everything works fine, the other point of the recognition of the objects (recognized earlier)stops (of the same window).