HeRMeS - cannot spy the object.

[AnatoliyBanniko]

Hi, I am now working on a project in HeRMeS, Oracle People soft, browser-based. I faced the problem of failing to spy the input field in a pop-up window in there. Pop-up is not of win32 nature, actually, I cannot understand what nature does it have. Once I spy it with AA, BP cannot see this field at runtime. And yes, I did avoid any dynamic parameters in AA for this field. I am also trying surface automation. I did make a screenshot of the desired pop-up window using BP and saved it on the disk. Then I take a picture of the whole screen I have during runtime using region mode. After that, by utilizing Find Image action in Utility - Image search VBO, I am searching for this pop-up on the current screen. Consequently, by finding X and Y coordinates of this picture on the screen I can adjust my Global Mouse Click aim. Yesterday during test runs in BO, BP was finding pop-up no matter what size IE is at runtime. However, when I tried to test this BO in a process, BP failed to find the desired picture of a pop-up window. Tolerance during image search is 100. I am desperate. Can anyone give me a hint? Thank you in advance.   BR, Anatoliy.

[Denis__Dennehy]

Something working in studio but not ran from the process/control room usually indicates your wait stage logic is not correct.  What logic are you using to wait for the pop up image to exist?  

[AnatoliyBanniko]

I think you did misunderstand me. Question is not about wait stages or control rooms. I am having all necessary wait-for stages each time screen updates, to increase the speed and performance.  However! The point I was trying to make is that - during spying [some]field has one set of parameters that help BP to identify it but when BP arrives at this page during another run parameters change in a way that BP cannot ""see"" them anymore. (As I said, I have omitted the use of dynamic params, such as links etc.) The same situation appears with surface automation. On one hand, trying to find the same picture in the same situation on the same screen should be an easy thing to do, keeping in mind pretty high tolerance(!) - 100 and BP completed this process perfectly during BO testing. On the other hand, during process runs in the studio, BP struggles to find this picture, therefore, fails. There are no obvious discrepancies in images, like at all. And if I would try to compare whole screen screenshot with the desired picture in MATLAB using image processing techniques, I would get some insignificant discrepancies that are obviously tackled by the tolerance! As I now know, the website is written using JS, although spying in Java mode - does not reveal any elements.  I have tried another way, though. Stupid, straight, can be unstable, but seems to be working. I spied pop-up with hidden HTML parameters that appear only when this pop-up is present on the screen. Then BP presses {TAB} several times to navigate to the desired field. Then search. Then press again {TAB} several dozen times and send {ENTER} - Hurray, completed!  :^)

[Denis__Dennehy]

I recommend looking very closely at what is different each time. For attributes, compare elements identified at different times to see what is changing and then change your match logic appropriately.  For image comparisons - compare closely in image app (paint etc) the original image and images that do not work - what is different/changed.  If your image includes text, is font smoothing turned off (is the text blurred slightly when you zoom in).  If your image contains text and is in IE - it may not be possible to turn off Font Smoothing within that application.    It is very rare for tolerance to need to be 100 or more, 40 usually works.  Compare colors in paint to see how different they actually are and if that really is your issue.

[AnatoliyBanniko]

Thanks for hints!