Some Version of a Process

When you're a large company, like Oracle, you can force your customers to do things your way. "Because we said so," is something a company like that can get away with. Conversely, a small company is more restricted- you have to work hard to keep your customers happy.

When Doreen joined Initech, they were a small company with a long history and not too many customers. In the interests of keeping those customers happy, each customer got their own custom build of the software, with features tailored to their specific needs. So, Initrode was on "INITRODE.9.1", while the Soggy Beans coffee shop chain was on "SOGGY.5.2". Managing those versions was a pain, but it was Doreen's boss, Elliot, who ensured that pain escalated to anguish.

Elliot was the one who laid out their software development and source control processes. It was the Official Process™, and Elliot was the owner of the Official Process™. The Official Process™ was the most elegant solution Elliot could imagine: each version lived in its own independent Subversion repository. Changes were synced between those repositories manually. Releases were also manual, and rare. Automated testing was non-existent..

Upper management may not have understood the problems that created, but they knew that their organization was slow to release new features, and that customers were getting frustrated with poor response times to bugs and feature requests. So they went to the list of buzzwords and started pushing for "Agile" and "DevOps" and "Continuous Delivery".

Suddenly, Doreen and the other developers were given a voice. They pushed to adopt Git, over Subversion. "I've looked into this," Elliot said, "and it looks like Git uses GitHub and stores our code off-site. I don't trust things that are off-site. I want our code stored here!"

"No, you don't have to use GitHub," Doreen explained. "We can host our own server- I've been playing around with GitLab, which I think will fit our needs well."

Elliot grumbled and wandered off.

Doreen took a few hours to configure up a GitLab instance, and migrate their many versions of the same code into something approaching a sane branching structure. It'd be a lot of work before the history actually made any sense, but it allowed her to show off some of the benefits, like that it would build and run the handful of unit tests she whipped up on commits to certain branches.

"That's fine," Elliot said, "but where's the code?"

"What… do you mean? It's right here."

"That's the code for Soggy Beans, where's the Initrode version?" Elliot demanded.

Doreen switched branches. "Right here."

"But where did the Soggy Beans version go?!" Elliot was getting angry.

"I… don't understand? It's stored in Git. We're just changing branches."

"I don't like this magical nonsense. I want to see our code in folders, as files, not this invisible shapeshifting stuff! I don't want our code where I can't see it!"

Doreen attempted to explain what branches were, about how Git stored files and tracked versions, but Elliot was already storming off to raise hell with the one upper manager who still listened to him. And a few days later, Elliot came back with a plan.

"So, since we're migrating to Git," Elliot explained to the team, "that poses a few challenges, in terms of the features it lacks. So I've written a script that will supplement it."

The script in question enumerated all the branches and tags in the repository, checked each one out in turn then copied it to another folder. "Once you've run this, you can navigate to the correct folder and make your changes there. If you need to make changes that impact multiple customers, you can repat those changes on each folder. Then you can run this second script, which will copy the changed folders back to the repository and commit it." This was also how code would be deployed: explode the repository out into folders, and copy the appropriate folder to the server.

At first, Doreen figured she could just ignore the script and do things the correct way. But there were a few problems with that. First, Elliot's script created commits that made it look like every file had been changed on every commit, making history meaningless. Second, it required you to be very precise about which branches/versions you were working on, and it was easy to make a mistake and commit changes from one branch into another, which was a mistake Elliot made frequently. He blamed Git for this, obviously.

But third, and most significantly: Elliot's script wasn't a suggestion. It was the Official Process™, and every developer was required to use it. Oh, you could try and "cheat", but your commits would be clean, clear, and comprehensible, which was a dead giveaway that you weren't following the Official Process™.

Doreen left the company a short time later. As far as anyone knows, Elliot still uses his Official Process™.

[Advertisement] Otter - Provision your servers automatically without ever needing to log-in to a command prompt. Get started today!

This post originally appeared on The Daily WTF.

Leave a Reply

Your email address will not be published.