what's everyone's opinion on when is it better to code your own scripts vs using recording? In my experience, it seems like manually coding the scripts usually takes longer, but if you have many similar cases, it's more reusable that way. And sometimes you need to do it to test for dynamic data. Am I correct? Or is there other circumstances that it's better to manually code the scripts?
4- While using robotJ for HTML applications, I found doing coding is far better than recording.
As for HTML object map is very much hierarchical and gives weight to Url and href properties too.
This URL and href gets changed in JSP's as par data loaded also. So multiple recording through same page makes same pages and objects recoginzed as different objects.
So i belivev in "record once, wrap in methods and used anywhere".
Secondly for data driven scripts coding is far better than recording many times.
You capture the object map in the beginning and then manually code the rest, is that right? Do you get a lot of "Weak object recognition" warnings from your scripts though? Are there any tricks in terms of minimizing these type of warnings?
Another questions I have is that how do you guys organize your tests to make them more reusable? Do you have objects representing inputs from each page? Or do you have individual cases for each input field? or some other ways?
Yes You are right, we create object map first.
Week recognition we get but that can be solved by adjusting the thershold if it is coming also it does not have any impact.
The bigger problem is to maintaining object map.
As if the same page and its objects are stored in object map more than once just because of change in some recogintion property than it will be problem if in future your UI gets changed.
And the whole idea of centralized change in case of UI object gets changes, will require more effort.
For ecah individual field will be a burdon.
What i do, I make application specific function based upon the application functionality ( including objects that will be used together ).
It may lead to us to grouping the Objects of one page or across pages or some times only one object.
Actually When i do this i think i am not able to take benifit of Object Map completly.
Some times while writing these functions i think if this object map would be more flexible to configure the properties for recognition , our life may be more easy :-)