State of the Market Software Testing

[eek]

Not just India but also the Phillipines.



Why should he? He's absolutely spot on.



You're not a pure test analyst if you have developer skills. And a developer shouldnt be testing their own work after unit testing anyway!

So you want to manually run the same regressions tests on every release.

Automated testing saves hours of monotonous work which could be better used testing of new functionality or edge cases

[BolshieBastard]

So you want to manually run the same regressions tests on every release.

Automated testing saves hours of monotonous work which could be better used testing of new functionality or edge cases

Small businesses dont see the cost benefit paying for an automated tool to run regressions. So they tend to run manual packs.

Every Enterprise company and the departments I've worked with (thats all the big banks, card issuers etc) have rarely invested in an automation tool to run regression.

Regression only forms a small part of the general testing market.

[xing]

You're not a pure test analyst if you have developer skills. And a developer shouldnt be testing their own work after unit testing anyway!

This is so out dated. Developers should be responsible for test their own code, and are expected to implement both unit and integration tests to ensure delivery of WORKING code. Simply throwing half baked code to testers is very irresponsible! The role of a test analyst is now limited to exploratory testing, while bugs discovered are handed to developers to convert to automated test cases. Very few of them are willing to pay for human to execute test scripts and tick spreadsheets!

[BolshieBastard]

This is so out dated. Developers should be responsible for test their own code, and are expected to implement both unit and integration tests to ensure delivery of WORKING code. Simply throwing half baked code to testers is very irresponsible! The role of a test analyst is now limited to exploratory testing, while bugs discovered are handed to developers to convert to automated test cases. Very few of them are willing to pay for human to execute test scripts and tick spreadsheets!

Re read the post. Developers are responsible for unit testing. They may perform a confidence test of integrated code but they should never be responsible for signing off Integration testing of their own code.

It goes without saying 'throwing half baked code' as anyone is foolish in the extreme (but many developers still seem to do this!)

[xing]

Re read the post. Developers are responsible for unit testing. They may perform a confidence test of integrated code but they should never be responsible for signing off Integration testing of their own code.

It goes without saying 'throwing half baked code' as anyone is foolish in the extreme (but many developers still seem to do this!)

Extreme? That's the very reality! Quite a few places run development teams this way.

At the places not maturing enough, the tests are implemented by a group of people called "developer in test".

[SussexSeagull]

I have been in testing for nearly 20 years and automated testing has always been on the verge of putting us all out of work. In the current climate most clients just want projects finished with the minimum of delay and barely want to pay for testing, let alone automated testing.

If you are implementing a new system that you expect to last a decade, there is a very good argument for automating tests but if you don't expect to do many updates there would be no ROI. Also, if a proper Test Analyst doesn't design the tests to start with it can be a waste of time.

Don't get me wrong, automated testing is powerful in the right circumstances but it isn't the cure all some people make out.

[mudskipper]

So you want to manually run the same regressions tests on every release.

Automated testing saves hours of invoiceable work which could be better used testing of new functionality or edge cases

FTFY

[SpontaneousOrder]

I have been in testing for nearly 20 years and automated testing has always been on the verge of putting us all out of work. In the current climate most clients just want projects finished with the minimum of delay and barely want to pay for testing, let alone automated testing.

If you are implementing a new system that you expect to last a decade, there is a very good argument for automating tests but if you don't expect to do many updates there would be no ROI. Also, if a proper Test Analyst doesn't design the tests to start with it can be a waste of time.

Don't get me wrong, automated testing is powerful in the right circumstances but it isn't the cure all some people make out.

The thing is that automated testing shouldn't really cost more than manual testing for anything non trivial. That's kind if the while point - it's cheaper, otherwise you'd just hire more testers.

If you're not starting from scratch then it can be a pain, but with a minimal investment in bootstrapping the whole process then it's not normally much additional effort for a developer/tester to knock up some fixtures or whatever for the new feature they're implementing. I'm writing unit tests anyway, and I'm going to manually test it before I throw it over the wall - so we might as well cosy up together and write a few fixtures and acceptance tests together, and then the tester can think up some exploratory testing with 90% of what he needs to automate it already in place.