Forum Discussion

RUDOLF_BOTHMA's avatar
RUDOLF_BOTHMA
Community Hero
6 years ago

aqObject.GetPropertyValue object does not exist after an exist check passes

Hi all,   New one that suddenly started and intermittently pops up.   Code:   var loadingPanelObj = Aliases.browser.FindChildEx(["ObjectType","ObjectIdentifier"],["Table","ASPxLoadingPane...
IE
Object Recognition
  • AlexKaras's avatar
    AlexKaras
    6 years ago

    Hi,

     

    > In theory if the .Exists is true, the next step shouldn't say it doesn't exist ?

    Correct. But only in case if the object was not recreated (or just destroyed).

    Note, that you are using Aliases to reference test objects and this is fine. But Aliases functionality has some specific that it caches objects once found and tries to reuse them later. And if the object is recreated, the problem will occur. See https://support.smartbear.com/testcomplete/docs/reference/test-objects/members/common-for-all/refreshmappinginfo-method.html for some more description.

     

    My guess is like yours: the target table might be recreated by script code on the page and this invalidates its previously found (and cached) instance. If this is the case, then you need to figure out the way (or ask developers to implement some flag for you) to identify if test code must wait for the table to be recreated.

     

    P.S.

    Out of curiosity: why instead of

    var isVisible = aqObject.GetPropertyValue(loadingPanelObj,"VisibleOnScreen");

    not to use just

    var isVisible = loadingPanelObj.VisibleOnScreen;

    ?

RUDOLF_BOTHMA's avatar
RUDOLF_BOTHMA
Community Hero

Edit: They actual respone to AlexKaras that got lost when my browser crashed:

 

I'm going to see if I get different results from Sys.browser rather than Aliases...

 

<

Out of curiosity: why instead of

var isVisible = aqObject.GetPropertyValue(loadingPanelObj,"VisibleOnScreen");

not to use just

var isVisible = loadingPanelObj.VisibleOnScreen;

?

Because of copy, paste and inexperience when I originally wrote it  :smileytongue: 

For curiousity, this is the code that follows it.  This panel is AJAXy,  So I need to keep polling to see if it's visible.

 

var retries = 0;
    while(isVisible && retries < 100)
    {
      retries++;     
      if(retries==2)
        LogMessageVerbose("***** WaitLoadingPanel() had to wait *****"); 
      if(loadingPanelObj.WaitProperty("VisibleOnScreen",false,100))
      {
        LogMessageVerbose("***** WaitLoadingPanel() finished waiting *****"); 
        break;
      }
    }

I could probably get away without the while loop, but it allowas me to log whether the load panel was visible and how long.

 

Regards,

 

RUDOLF_BOTHMA's avatar
RUDOLF_BOTHMA
Community Hero
6 years ago
I'm going to see if I get different results from Sys.browser rather than Aliases...

 So far so good...

 

Now to figure out why I end up with mutiple iexplorer.exe processes running (and edge as well) after a while.  At some point IE just starts a new process and I get ambiguous browser.  Thats for a different thread though

 

  • AlexKaras's avatar
    AlexKaras
    Champion Level 3
    6 years ago

    > At some point IE just starts a new process 

    Yes, I've also have met this on some environments.

    It looked like that IE process was hanging somewhere on close (though this could be due to the tested web application) and any attempt to address it from TestComplete (like browser.Exists, etc.) caused its crash or TestComplete problems. As the chances for the problem to be investigated by developers were extremely low (the problem could not be reproduced manually), we ended up with the approach to .Terminate() the browser process if tests were executed against IE browser.

     

  • RUDOLF_BOTHMA's avatar
    RUDOLF_BOTHMA
    Community Hero
    6 years ago
    RUDOLF_BOTHMA wrote:
    I'm going to see if I get different results from Sys.browser rather than Aliases...

     Yes, it looks reliable combined with browser wait.  I don't want to use RefreshMappingInfo if I can help it.  This script gets called on virtually every KWT, so refreshmappinginfo() would push the test time up by a huge amount.  There are places where NameMapping is extremely useful because sometimes identification is a mess and naming too, but I may have to work this into the mix to avoid the caching issues I have experienced previously and had to work around by refreshing.  Could end up with nice performance at the cost of code readability.  Some object names just don't tell you anything about what you are working with :smileysurprised: