Software Testing

I came across this comment on Slashdot yesterday:

Top flight developers producing quality code don't need large QA departments. They've already written well-designed, bug-resistant code, unit tests, integration tests, and performance tests, all in the course of producing something that works (the first time). If you have to pay a phalanx of QA engineers to find bugs post-facto ("just as important as our development department"), you're doing it wrong. The bugs shouldn't have been there to begin with.

Anyone who has ever done any serious software development knows this is not true so I'm assuming that either the poster was a troll or isn't a software developer -- most likely a troll.

Unfortunately the false nature of this statement was well illustrated this morning in my new TagME project. I do a lot of different things to try and catch bugs as I code but there is no replacement for human testing. When developing the export to KMZ feature I hard coded the output file to be output.kmz in the current directory and forgot to change this and allow the user to choose their own file and directory before I released it last week.

Fortunately, releasing TagME as a Web Start application means that every time anyone starts it up, either from the website or from the shortcuts it installs, it automatically updates itself to the newest version. So I've recompiled everything and placed a new version online so next time anyone launches TagME they will get the bug fix without knowing. They will know something has happened though as the new version also includes some new features. I've restricted a lot of the input fields so they now only take valid values which means it is a lot harder to crash. Also you can now specify the colour, thickness and opacity of the track line, and change the orientation of the photos. I'll update the tutorial later today.