I'm having a weird issue with text recognition. When I start up Winrunner, it usually works fine at first but stops working after a while. Any call to a get_text function returns nothing. I tried with both driver and image based text recognition. The only way I found to fix this issue is to log off from the workstation log back in. Did anyone ever experienced a similar issue?
The software we're testing is developed by a third party. If we want them to build the required Delphi unit into their executable, we'll have to write a change request pay extra money. I work for a very large company and this process is long and complex. The Delhpi add-in is something we're considering but we'd like to get something done before the process goes through.
Is there a relationship between text recognition and add-ins? Shouldn't WR just "convert" the window into a bitmap and return whatever it recognize (just like an OCR system would do)?
There is a relationship between object recognition and attribute information and the add ins. It's possible that the text associated with an object (as a label) exists as an extended, non standard attribute which would be available as part of the addin.
On the OCR side of the house, you do have the ability to "educate" WinRunner about different fonts and font families for recognition, but if WinRunner only sees a box, but not the contents of the box due to an inability to communicate with a native delphi component, well, you might just have to get by with an expected result which is a bitmap (and less than ideal...)