Hi,I just whant to clarify that I am using system full path which mapped to an object and not using name mapping option at all.Some time my scripts get fail due to object identification issue.Does its a best appraoch or need to use the name mapping in order to create scripts, which not falling. .............. Thanks Dev

I set-up a script that acts as an "object repository" and define all the objects as functions, like this:var testBrowserType = "iexplore";function mpPortal_Link_FAQs(){return WaitForAndReturnTheObject(Sys.Browser(testBrowserType).Page("*"), new Array("ObjectType", "innerText"), new Array("*Link*", "*FAQs*"), 99);} //wait 5 seconds if page and object load time doesn't sync properly with page.Wait() var objWaitTime = 5; function WaitForAndReturnTheObject(page, attribs, vals, depth, waitTime) { var object; var theWait = 0; if(waitTime != null) theWait = waitTime; else theWait = objWaitTime; for(var i = 0; i <= theWait; i++) { object = page.Find(attribs, vals, depth); if(object.Exists) { return object; } else if(theWait > 0) Delay(1000); }//end for try { Log.Error("Object not found! " + attribs[0] + ": " + vals[0]);// + ", " + attribs[1] + ": " + vals[1]); } catch(e){} return object; } //end WaitForAndReturnPageObject(page, attribs, vals)

In my experience...neither. I try to find everything dynamically by it's name, text, or caption. In some applications I use text recognition. The reason I find everything dynamically is probably a common issue for many people where if any parent object in the hierarchy changes then every child object will need it's name remapped or fixed. If you have large scripts this can be an EPIC undertaking. I suggest looking at .Find, and .FindAll methods as well as text recognition.

Unless you use extended find in your name map.But that comes with it's own pitfalls. Namely, performance hit (it slows things down) and also that by using it you can *possibly* be masking a breakage you may actually be interested in finding. I guess it's also possible that it could actually find the wrong thing, but that hasn't happened to me yet.I use a combination of both. Name maps (with Aliases - full name map reference are pretty horrible) for the parts less prone to changing at run-time and on-the-fly identification for the more dynamic parts.Either way, if your application is still subject to change, you're likely to hit problems of some sort at some point along the line. Unless you do all your searching from the base level of the object tree .... which I would imagine would also cause performance issues?

Hmm, I'm not as familiar with pure java based applicatioins for TC; our application is driven by the browser. Can SWTObject's not be defined with more specific "named" properties? One thing I've found in my software development work is that developers are usually amenable to helping automation testers by giving good name mapping properties to page objects, if you catch it and ask them during initial development at least.

For dynamic sites/apps NameMapping is evil! You will have to do lots of maintenance :)In other hand, for static products is a great feature, it depends on your product.I always search for objects as described by Ryan.

You might also want to consider how to handle the name mapping file if you have concurrent test development on your project and or source control. With multiple users you'll have to manage and merge different versions of the name mapping file. I use the same approach as Ryan and have a script that locates the object using the find method and it works just fine. Way more robust as we only need a property name and value pair to locate any object on the page and not have to care about the path to the object.

What is the best approach for scripting with name mapping or without name mapping?

16 Replies

amathews
Occasional Contributor
10 years ago
I set-up a script that acts as an "object repository" and define all the objects as functions, like this:

var testBrowserType = "iexplore";

function mpPortal_Link_FAQs(){return WaitForAndReturnTheObject(Sys.Browser(testBrowserType).Page("*"), new Array("ObjectType", "innerText"), new Array("*Link*", "*FAQs*"), 99);}

//wait 5 seconds if page and object load time doesn't sync properly with page.Wait()

var objWaitTime = 5;

function WaitForAndReturnTheObject(page, attribs, vals, depth, waitTime)

{

var object;

var theWait = 0;

if(waitTime != null)

theWait = waitTime;

else

theWait = objWaitTime;

for(var i = 0; i <= theWait; i++)

{

object = page.Find(attribs, vals, depth);

if(object.Exists)

{

return object;

}

else

if(theWait > 0) Delay(1000);

}//end for

try

{

Log.Error("Object not found! " + attribs[0] + ": " + vals[0]);// + ", " + attribs[1] + ": " + vals[1]);

}

catch(e){}

return object;

} //end WaitForAndReturnPageObject(page, attribs, vals)
googleid_118035
Contributor
10 years ago
Hi All,

Thanks again for your comments.

......................

Thanks

Dev
mtsmith
Contributor
10 years ago
This has been a great discussion on name mapping experiences, practices and pitfalls. It would be useful to more know about the types of applications being tested and the most reliable way to approach each type. Our applications under test use Java SWT and this presents a number of challenges since map able objects rarely contain useful properties for creating aliases. As an example, if there are 10 text boxes on a screen, they will be given the names SWTObject("Text", "", 1) ... SWTObject("Text", "", 10) and it's up to the user to assign the aliases. Then, if on the next application release another text box is inserted between ("Text", "", 1) and ("Text", "", 2), all text boxes from #2 onward will have aliases corresponding to the wrong text box. This is true for most SWT editable controls and it limits the use of TestComplete's aliases.

As a workaround, a mapping system has been built that first attempts to associate controls with the nearest horizontal or vertical label. The best matching information is saved in a custom application map stored as XML and this map used when running the test scripts. It's not always possible to find a valid label/control match, so a variety of other methods are also be used. In cases where controls such as text boxes are arranged as grids, row labels and column labels are used to locate the text box at the screen location intersection. TextObjects have been very useful for identifying dynamically generated objects such as composites within expand bars. In rare cases, the Java data field has been used along with several other mapping functions.

Like many other posts, in some cases we've moved to using one of the variants of the find method to dynamically reference objects. It has proved to be the most stable. Also, each situation is different, so the solution with the best reliability and performance is used. What works well in one application may not be suitable in the next.

So, in our case, the aliases provided in TestComplete were not as reliable as desired and a custom mapping system was developed. I'm definitely interested in hearing comments from any Java SWT testers out there.
mtsmith
Contributor
10 years ago
That's the core of the problem, in Java SWT the identifying information the developer provides for most editable controls is not availble via the Java Runtime and TestComplete. What is available, is the type of object and an index.
amathews
Occasional Contributor
10 years ago

thePanel = parentPanel.FindAllChildren(new Array("ObjectType", "innerText"), new Array("Panel", "*" + textBoxIdentifier + "*"), 9999);

thePanel = new VBArray(thePanel).toArray();

foundRightPanel = false;

//Text directly preceding textbox with ONLY spaces and no text

//before it in the same panel

re = new RegExp("^\\s*" + textBoxIdentifier);

len = thePanel.length;

panel = null;

for(var i = 0; i < len && !foundRightPanel; i++)

{

if(aqObject.IsSupported(thePanel[thePanel.length - i - 1], "innerText"))

if(re.test(thePanel[thePanel.length - i - 1].innerText))

    {

     panel = thePanel[thePanel.length - i - 1];

     foundRightPanel = true;

    }

}

theTextbox = thePanel.Find("ObjectType", "TextBox", 9999);
leandropoblet
Frequent Contributor
10 years ago
I'm currently testing a WPF based application.

I use a combination of both methods: whenever I can use NameMappings I do, whenever I can't I use a function I made which receives the parent object (usually mapped name), property/ies and value/s and it handles zero, one and many objects found.

So I use as much as I can the NameMappings.

Cheers,

Leandro

Forum Discussion

What is the best approach for scripting with name mapping or without name mapping?

16 Replies

Related Content

Reusable REST Request approach

Mapping lost

Script Extensions that require Name Mapping files

Mapping TestComplete scripts against VB6 application. Losing mappings very rapidly.

Mapping license for testexecute

Recent Discussions

Best way to handle Azure login using TestComplete

Unable to access Navigation menu component sub menus

Unbinding a project suite from source control

Is it possible to test editing a pdf form in a browser?

how to use Random Data every time you run a TC