criteria of automation for a test
Which are the criteria allowing to know that a manual test is automatizable?
It is very significant to use an automat in certain case. But how to decide that a manual test is automatizable?
Is there any document to help us to define if we can lose time to use the automaton to create a script?
(an exhaustive document.)
Too much company wastes time with tried to automate functionalities that are too complex to automate.
Re: criteria of automation for a test
When to automate or not ?????
Its simple economics. Test tools and specialists are expensive step one is estiamte the extra cost in automating. If this figure is way too high then that is the end of the analysis.
The next step is to work out the life cycle of your product. If the software is short lived then forget automation altogether.
If you have made it to here then you need to look at issues such as the development cycle and how much of the code base is to change between releases. My personal view is that if 15% or more of the code base is to change between releases....forget automation. You will spend more time maintaining scripts than actually testing.
Finally there is the case where Automation is the only way to test a particular event. Some tests human operation maybe too slow or clumsy to ensure test is conducted properly.
The value of Automated testing is inherrent in regression testing. One has to determine the cost of maintaining an Automated framework (licences, training and specialists) versus hiring a bunch of manual testers. The longer the life cylce and the bigger the project the more likely an automated test tool can add value.
[This message has been edited by rtehve (edited 04-14-2003).]