Hi,I am a new memeber to this forum.Nice to see you all here!
Our company has been doing many kinds of manual testing.Our engineering team is well versed with using Rational products like Rose and Clearquest.We are planning to Automate our manual testing of our Webasites using WinRunner.Your inputs on..
1.What are ways in which the Automated tool"WinRunner" can help "Webtesting"?
2.What are the typical "Checkpoints" used in Webtesting context?
will really help me!
Interesting your not doing Robot and test factory.
1) Coming from a Rational group, is this a trick question?
WebTest is a library of routines tailored to run IE or Netscape environments. WinRunner is actually a generic window test tool (set) for any windows application. If your web only, try Astra Quicktest lessons and demo download (for a week) are free. Take the lessons first, then load, it makes for a more progressive evaluation week. Its easier then WinRunner (and may be better).
Tests (automated or manual) are designed to meet objectives, usually requirements. Automated tests can be broken down into a few classic sets. The rule of thumb is that a test must be conducted 2-3 times before the cost of construction can be recovered from savings over manual testing. In other words, one time projects generally can not be cost justified.
Classic sets (to me anyway);
Requirement Does it work?
Regression Did we break anything?
Performance Does it handle 200 transactions an hour?
Start over - We changed it so start over!
Start over, was the best-cost recovery I personally saw. Development makes index or table changes in the database but rather then convert old data they just gave us a blank one. The application front end had little or no changes. Using automated routines that read excel spreadsheets to fill out the entry forms we had the database populated in a half day. The team of 6+ QA could then continue work on things like summary reports.
Anytime you need to wait for an event (IO, timeout), wait and check that it finished. I prefer window names, text, and bitmaps in that order. You usually only need to check one thing to know it completed correctly. I avoid wait() calls if I can, I try to find an object that signals completion. Say I wait 10 seconds every transaction because the test system is sometimes slow. Even in production I end up waiting 10 seconds. If I am event driven, it runs as fast as it can regardless of where.
PS I have used Rational Test Suite, mostly Requisite Pro but worked with the other parts. I was a member of the local user group. For those who have not studied RUP Do it! They have some of the clearest thoughts in the business. (Free at Rational web site.) It will help you, even using Mercury tools.
QA Architect - Software Quality Practice
4411 S Cross St, Downers Grove, Il 60515