2017 m. kovo 8 d., trečiadienis

Before release do regression testing

   Ah, regression, my worst friend, the biggest foe. Rarely there is a product in production without having occasional or regular regressions. Believe me or not, I did several regression testings over my career as tester. If You are thinking - "oh, he had just several, lucky bastard", well, You are wrong, had a lot, can't even remember all shit I have tested, and how many times...

   For those who doesn't know what regressions are (You are lucky) - this is written down test cases with steps and expected result, and most of time there is one test case for every action, event or stuff which is important to applications work, and it's need to be performed before product release. So there are a lot, hundreds, of tests which needs to be run over in specific period of time, before new version of application can be released in production. Definitely You can find different descriptions, but overall everything ends kinda similarly.

   How to deal with regression? If possible outsource it, pay for other companies to do monotonic work for You, and You can focus on more interesting stuff, like.. I don't know, video games or something. Other possibility - automation. If You can, automate everything. Add every new bug and story to automated cases and forgot about it until You find it crashing, fix it and forget about it again. And of course, funniest part - DIY.

   Did Your knife ever become dull?

   Knife is working - cutting through stuff, scratching, carving on doing other things which is designed or not to do. Some small bends appears on edge, it becomes dull, and maybe other things were not finished from last release. And one day comes, when it needs to be sharp again for putting it back to work. Same as software before release. Bugs where found, stuff needs to be updated, new features implemented, so knife needs to be sharpened.

   And you need to do it Yourself, because You are cheep and don't want to pay to other company to do it, and You don't have some kind of fancy electric hone. Doing it at least one time with our own hands will open Your eyes for stuff which were missed, skipped last time, and how poorly implemented this time.

   I have nice fancy water stone, it is not simple rock, it is designed to be used for honing. You shouldn't be using stuff which is not designed to it, You can, for example, have all Your test cases in excel spread sheet, but there is better tools precisely for that. If You aren't making it automated, at least make it easier do it manually. 

   Doing it once it will increase You understanding. Doing it ten times, You will get good at it. And after couple of years, You will want automation or outsource it as hell. Well this is my experience, maybe Yours are different, would be nice to know. Or not, Your choice.

   No funny image this time, because regressions aren't funny.


Your friendly neighborhood Tester


Komentarų nėra:

Rašyti komentarą