namemapping - mapping child

Hmm. Not specifically.



But I know a couple of the people on here make extensive use of them:



https://smartbear.com/account/profile?userid=27549 (Ryan Moran)

and

https://smartbear.com/account/profile?userid=8817 (Jose Pita)



It came up in a thread a while back ...



http://smartbear.com/forums/f74/t91186/core-layout-change-effected-dom-most-items-are/#91215



And one of Jose's posts links to:



https://gist.github.com/jpita/9954138



Which outlines a lot of the helper functions he uses.



I know Ryan uses central ones which he passes a top level object, and some identification properties to, which then searches the object tree below it looking for the object in question using the properties supplied and series of filters. But you need to exercise some level of control. If you search from the top of the tree for every single object, I can't see that being good for performance.



When you use such functions is another thing you have to factor in. You can simply search out objects on the fly during the run. Or you can have a "setup" function which searches out all the objects the run will require in advance and stores the objects for use during the run. The second approach requires more work and forethought, but would probably have less of a performance impact during the actual test part of the run.



I make some use of such things. But I also do tend to use name maps (as you'll see from the thread I linked to above) unless something is hugely dynamic. In those cases, I have used helper functions but they tend to be fairly application specific (so far, mine have had to use combinations of properties of several sub elements, which return a higher level element, which is then stored and has further tests run against it and it's child objects) so wouldn't be much help to you beyond the concept I've just outlined.



Object identification in automated testing is always a hot potato and, as far as I'm concerned anyway, I've yet to find an answer I'd be happy to use in every single scenario.

If it's a highly dynamic application (which it sounds like it is) you may be better off using runtime helper/finder functions and finding everything on the fly during a run.



The more dynamic something is, the trickier it is to make it work reliably with name maps.

Hallo,

Thanks Ann for the answer. I knew of the "map child" option: what I was asking (sorry, I was not clear) is it is possible to map child recursively, so not only the child object, but also their gran-child and so on.

Hi,

As far as I know, there is no such functionality in TestComplete. We have an existing feature request for that, and I've increased its rating based on your post. Thanks for the feedback.

By the way, what do you need to map such a great number of objects for? Could you please describe your testing needs? Perhaps, we can find an alternative solution for you.



Ann

it is just to avoid conflict while working with many testers. If you create the whole namemapping at the beginning, there are less chance of having conflicts. consider that most ofthe tester which want to use test complete will just record what they normally do manually, so they won´t modify the namemapping in any way.

Hi Lorenzo,

Thanks for the explanation. As a workaround, you can enable the Map object names automatically option and record a dummy test. While recording, click over all the objects and open all the windows that you plan to test then. TestComplete will map them automatically. After that, you can delete the recorded test and use the created Name Mapping scheme in your tests.



Ann

Hi,



Realising this is an old post I would still like to comment. I am quite new to the forums so please bear with me.



The ability to recursively map child objects would be extremely helpful to me as well. We develop java apllications. Two versions of these applications are compiled, a version with debugging enabled and one without. This causes the base prosess name to different for both applications even though the rest of the object mapping is identical.



I have created tests that will work on both versions without needing to rewrite the tests for the different version but for this to work effectively I have to map both versions manually before compiling the tests. This is quite time consuming but effectively saves a lot of time debugging. 



I would find it extremely helpful to be able to recursively map the objects, even if this takes a while and only gives me default name mapping.



Carl

Hi Carl-Hartwig,

Hi Carl-Hartwig

 

We work in a similar environment whereby there are three base applications, so the process name is different, but thereafter, all the windows, controls, etc. are the same.  In order to avoid having to re-map each of these separately, I created three name mapping configurations, i.e. all of the controls have to be mapped once anyway, so when that is complete, use the name mapping editor/configuration manager to copy it to a new configuration.  So, at this stage I have 3 different configurations, each with their own name which corresponds to the version of the application that they are for.  So, when I have to test any one of them, the first thing my tests do is to call a shared method, which determines what application is being used and applies the desired name mapping file.  Is this something that might work for you?

 

You can find out further details in the on-line help, under Name Mapping >> Configurations.

 

Regards

Stephen.