Tests? We don't need no stinkin' tests!

[Jetfire]

It can be difficult to do. I joined my current company about 18 years ago, and they didn't really have a Unit Test framework at the time. My job before it was heavy on Unit Testing and I did miss that framework at the time, but I was too junior to push it back then. Still a few years later, some of our teams started developing unit tests and making it our policy. There was some resistance, and it's taken a long time to get decent coverage. But now, everyone is grateful we have the coverage we have for the tests. It has allowed us the freedom and confidence to do some major refactoring and updating of our systems and know we haven't broken things worst than it was before.

At the moment the code base I work in the most has about 25,000+ unit tests; takes about 5 minutes to run fully without debugging in release. And some of our other big modules have 10k tests as well; all written in the past dozen years or so.

[Ironclad Alibi]

Quoth Nunavut Pants

There are quite a few people who agree that it is a fabulous idea, and that Somebody Else should put a whole lot of time and effort into developing it.

The problem with that approach is that they usually want that Somebody Else to pay for it.

[Nunavut Pants]

Quoth Jetfire

... But now, everyone is grateful we have the coverage we have for the tests. It has allowed us the freedom and confidence to do some major refactoring and updating of our systems and know we haven't broken things worst than it was before.

"The key to fearless refactoring is good test coverage."

Sounds like your current job is much like my first group at $(MAJOR_NETWORKING_COMPANY). We had good coverage, support from management to require people to test, and good people doing the work.

... so of course the mandate came down from Very Upper Management that we had to increase our release cadence and reduce head count by half. The top guy in our $(BUSINESS_UNIT) told them it couldn't be done, so they just closed down $(BUSINESS_UNIT) and got out of the $(PRODUCT) market all together.

[mjr]

Quoth Nunavut Pants

There are quite a few people who agree that it is a fabulous idea, and that Somebody Else should put a whole lot of time and effort into developing it.

I tried to be that Someone Else for a little while, but wound up unable to put together anything useful...

There are lots of useful websites and such as to unit testing online. I know it's in C#, but one book I'd recommend is Roy Osherove's "The Art of Unit Testing". There are plenty of others, too.

[otakuneko]

Going along with the testing theme--

My company recently released a new version of our agent which was supposed to support a specific version of Hive. A very big and influential customer has been waiting for this for quite some time. You can probably guess where this is going.

It didn't work. There were multiple issues with it, including in some cases crashing the server, but I dealt with one that was a real beauty. No crash, but it just plain didn't work. Threw a bunch of errors and sat in the corner pouting.

The problem? Oh, there was a code change in Hive, wherein a big B was changed to a little b. No big deal. It only broke our agent, which was supposed to support this specific version, which the customer had been waiting on for I don't even know how long.

Is it really a necro, when the thread is only #2 on the page?