Tuesday 15 April 2014

"if it hurts, do it more often."

I recently came across the phrase "if it hurts, do it more often."1 on the website of Martin Fowler.

Now, this might not be appropriate to, for example, slamming your hand into a car door, but it does have its uses, for example, in sports.

Personally, I think it is one of the defining characteristics of humans versus animals, that humans can suffer through a bad cause (pain/hardship/uncomfortableness) if they know the effect later is appropriately good. In other words, humans have the ability to reason about causality3, usually with a perception of time.

Unfortunately, sometimes the animal instincts prove stronger.

Continuous Integration

The soundbite in the title comes straight out of the realm of Continuous Integration2.

I came across the sentence, as I was wrestling with a problem at work. We have several branches, which are specific to certain of our customers. An often heard complaint nowadays is that these are so called 'Long-lived branches'. Now, the more I consider it, the more I think long-lived branches are a pretty bad idea.

The complaint I hear most often is that some of our software developers are spending more time merging changes to the different branches and keeping them in sync and retesting, than actually developing new software.

I just thought I'd put down some references (see [4] and [5]) on how to get rid of long-lived branches, and keep everything in the main branch. I don't feel sanguine about convincing management, though.

References

[1] Martin Fowler - Frequency Reduces Difficulty
http://martinfowler.com/bliki/FrequencyReducesDifficulty.html
[2] Wikipedia - Continuous Integration
http://en.wikipedia.org/wiki/Continuous_integration
[3] Causality
http://en.wikipedia.org/wiki/Causality
Lean into the pain
http://www.aaronsw.com/weblog/dalio
[4] Martin Fowler - Feature Branch
http://martinfowler.com/bliki/FeatureBranch.html
[5] Martin Fowler - Branch by Abstraction
http://martinfowler.com/bliki/BranchByAbstraction.html

No comments:

Post a Comment