automated tests - Regarding Automation Approach -

supposed automate application developed in powerbuilder. in order test application using rational robot functional testing tool. expect @ least 40 -50% of change control in application each release. release trends scheduled @ least 3 times in year.

the product has different setup each client. accordingly scenario has been derived. although if there change occurs, in functional feature , in interface. pointing that, need proceed automation. identified few areas stabilized (i.e., no major changes occurs) automate. feasible proceeding automation?

could please suggest me how go on this?

i've seen answer last question involve consultant spending 2 days interviewing development team, 3 days developing report. and, in cases, i'd report preliminary, introductory, , rushed. however, let me throw out few ideas may manage expectations on team.

testing automation great checking regressions in functionality allegedly isn't being touched. things framework changes or database changes can cause untouched code crash. risk-averse environments (e.g. banking, pharmaceutical prescriptions), investment in automation worth effort.

however, i've seen often underestimation of effort. test functionality in unit (let's window, example), need review specifications, design tests test each functional point, plan data (what data going window, how make sure data expected when start test each time, data when you've finished tests, how ensure data, including non-visible data, correct), script , debug tests. i'm not sure professional testers (i'm developer trade, have taken course in automated testing tool), if aren't planning same amount of effort expended on developing automation tests you're spending on application development same functionality, think you'll become frustrated. add changing functionality means changing testing scripts, , automated testing can become significant cost. (so, tell manager automated testing doesn't mean push button , things tested. < grin > )

that's not can't expend less effort on testing , some results, pay for. having script opens , closes windows in app provides some value, won't tell new behaviour implemented in framework being overridden on window x, or database change has screwed sequencing of items in drop down datawindow, or report completion time goes 5 seconds 5 hours. however, again, don't underestimate effort. new tool new language , idiosyncrasies need figured out , mastered.

automated testing can great investment. if cost of failure significant, badly prescribed drug causing death, investment worth it. however, cases there high turn on of functionality (like think you're describing) , consequences of failure less critical, might want consider comparing cost/benefit of testing automation additional manual testing resources.

good luck,

terry