Continuous Hipsterism

Software Engineering, Leadership, Management and more

There are no free meals

Posted at — Feb 20, 2016

Bob: “Hey, what will we use to communicate with X in this project?”

Ted: “We’re using library Y, of course! 5k stars at Github. Looks solid!”

(5 months later…)

Bob: “Ted, we have a problem. We need to update X and library Y does not have support for it.”

Ted: “Hmmm… We could just monkey-patch Y and then create our own in-house version. Problem solved!”

(and then it begins…)

Dear CTO/VP of Engineering / Head of Development, please do not allow your team to do these kind of things. They are bad. They are bad for your software, for software, in general, and they will hurt your company, sooner or later. Why? You may think that this is the quickest way to solve the above stated problem, but let’s talk about what could happen:

All of these can be heavily mitigated, if not totally, if you said to your team: “Go to Github, clone that project, create a branch, make a pull request and get your code reviewed. See? Seems easy, right? So why don’t you do it? Because you think that is quickest to do the “dirty quick-life-saver fix”, but down the road things can be much worse. Remember that guy that wrote the monkey-patch? Yeah, he’s working for another company now. And, of course, the library that your team patched now cannot be updated and take advantage of the new features that were implemented. Because, now, your code is entangled with n+1 monkey-patches to that same library. Yeah, that sucks. So, please, stop doing it.

Conclusions

Apart from the technical debt argument - hey, technical debt is as bad as the financial one - you probably owe the community. Your company was only allowed to bootstrap and to quickly develop the first features because you built it upon someone else’s work. You found a bug? You think you can improve it? The current maintainers do not have that much time and did not implement the feature you need? Contribute back. Go to the goddamn git repository, fork it, code it, test it and make a pull request. At the end, everyone will be happier. You’ll not end-up with your own in-house version, you will (probably) not break the library, you’ll get your code reviewed and you’ll be able to update the library in the future. Cool, right? :)