Iíve seen several posts in functional forum that describe following: number of test steps in test case (procedure) that is supposed for automation should be in general higher comparing to manual test case.
I wonder why? I know one reason why it should be just the opposite: you canít write so smart automated script that workaround any issue it run into. As a result single issue will cause whole scrip to stop (to be not executed). Manual tester is able to workaround most of the issues and continue testing.
?:the art of a constructive conflict perceived as a destructive diagnose.
The idea is not in every case to work around an issue. The idea is to find issues and communicate them! Granted, you may opt for, or be forced to work around known/documented issues in order to proceed with other testing.
I wouldn't expect a matchup of manual steps with automated steps since tools like WinRunner or Robot can bundle a number of validations into single functions.
Broken Jake has some very valid points, more validation points can be performed with automation than should be covered in manual tests, and it is next to impossible to tie automated steps to manual steps. Plus all tests pertaining to a specific functional point can be assembled into a test set or actually in a series of steps in one script, taking advantage of reduced setup procedures for each test where manual tests are run more randomly requiring setup for each test. That being said, the manual tests to be automated should be as simple and precise as possible with no more than one validation point per step and one main validation point to satisfy the requirement being tested.
Success is the ability to go from one failure to another with no loss of enthusiasm.
~ Winston Churchill ~