Page 1 of 2 12 LastLast
Results 1 to 10 of 11
  1. #1

    missed IEWindow Attach

    My script for testing a basic web application miss nearly 20% of the attachments of an IEWindow.
    If I look when the script runs, it suddenly seems TP can't recognize the window. If I stop the script and check if the object map can locate the window, it finds it without any problems.

    It's a wait statement where it occurs. It can't be the timeout because I have extended those without any better statistics.
    I tried to add pause,extra attach commands
    Is there anybody who have some experience with this behavior and how you solved it ?

    Regards Jorgen

  2. #2

    Re: missed IEWindow Attach

    did you try something generic such as "Application=iexplore.exe"? This should attach to any IE window in any state.

  3. #3

    Re: missed IEWindow Attach

    My application has an issue where GUI items are not visible to TP until a flash animation has fully loaded in an IE frame inside the app. (yes it's weird..)

    In your place, I would create a special error handler to dump the attachname of all currently displayed windows to a log the instant it occurs, maybe that will give you a clue as to what's going on.

    If you stop TP and then try to examine windows manually, thing could change before you get to it.

    Sorry it's not more help, but those problems are kind of tricky, and usually the cause makes no sense and it found in the funniest of places.

  4. #4

    Re: missed IEWindow Attach

    From having a objects very uniq. I tried to make them generic withou any luck. Yesterday I also added a pause before and it seems to be much more stable now. But the pause will add 1 second to the total measurement of the response time. i opened a case with CW also to see if this is a wellknown defects/"features".

    Thank you for your quick answers!

    Regards JOrgen

  5. #5
    Senior Member
    Join Date
    Apr 2003
    Wisconsin, USA

    Re: missed IEWindow Attach

    I have noticed that sometimes when I am doing a straight line recording session, moving from window to window, the initial window attach is generated, but subsequent window attach statements sometimes are not. In that case, I would simply get browser attach statements, but not the corresponding window attach statement. When I run the code back, sometimes the timing is such that TP gets "lost" because it still thinks the original window is up when a new one is where I want to be.

    In that case, what I do is stop the record every time a new window is displayed, then start the record again immediately. That way, all the proper window attach statements are generated.

    Take a look at your code and make sure that you have paired window attaches and browser attaches. If you don't, either re-record those parts to get the window attaches, or put them in by hand.

    Jorgen - I would be very wary of using TP for measuring response times. I sure wouldn't use them for any sort of performance test acceptance criteria. I say that, because in addition to application timing, you also have TP overhead involved. TP is not optimized to do performance testing.

    Also, take a look at this thread regarding timing. It may or may not be part of the answer.


  6. #6

    Re: missed IEWindow Attach

    I have noticed that whenever I create a logic event for "wait for exists" for a new window that the script will often fail at that point even though the new window appears.

    I believe the problem is that the window appears before the logic event triggers... I inserted a delay instead of the wait for event and never get a failure now.
    "I think when you spread the wealth around, it's good for everybody." -- Toledo, Ohio, Oct. 12, 2008

  7. #7

    Re: missed IEWindow Attach

    DSquared, can you explain further why you wouldn't use Test Partner for performance testing. Compuware's performance testing application Client Vantage uses Test Partner for all of it's scripting. I am a little concerned because I have converted my scripts from QA Run to TP and I have noticed that some of the timings have increased by a second or so. I have verified that I have the exact same events as I did in QA Run.

  8. #8

    Re: missed IEWindow Attach

    I can't speak for dsquared but i believe he already answered your question:

    "I say that, because in addition to application timing, you also have TP overhead involved. TP is not optimized to do performance testing."

    VB6 is not a very fast language and TP can be EXTREMELY slow when dealing with a large # of gui objects. It also lacks sophisticated timing mechanisms so you won't really get fine-grained performance data.

    I would trust TP to measure performance with an accuracy of within 5 or 10 seconds.. IF the GUI was simple.

  9. #9
    Senior Member
    Join Date
    Apr 2003
    Wisconsin, USA

    Re: missed IEWindow Attach

    TestPartner is also single-user. How would you ever be able to do multiple VUs with TP? It simply is not architected to be a performance testing tool. Never was meant to be.

    I suspect that the increase in time has something to do with the VBA overhead. But that is only speculation.

    QALoad is specifically built to do performance testing.

  10. #10

    Re: missed IEWindow Attach

    Jorgen, did you receive any feedback from CW?

    I have a suite of 310+ web tests, each one starts a new instance of IE after closing any existing browsers.

    Randomly get between 20 and 200 attach failures during test execution. The failures are on different tests each time, always at:


    Adding wait statements throughout to mitigate against synchronisation issues proved fruitless, as did using alternative generic attach property values.

    Going to try above suggestion from grantq to dump all open windows.


Page 1 of 2 12 LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
BetaSoft Inc.
All times are GMT -8. The time now is 05:53 PM.

Copyright BetaSoft Inc.