CodeReviews
Post #4829, on Aug 28, 2022 in TG
Developer desperation for PR approvals illustrated by cowboy meme with dubious negotiations
Description
Low-resolution, VHS-style frame of a shirtless man in a black cowboy hat standing beside a parked car outside a strip-mall convenience store. White subtitle text at the bottom reads: “-You’re prostituting yourself off for d■■■■ again, aren’t you? -A man’s gotta s■■■■ Code.” An additional white overlay near the car door labels the vehicle occupant as “PR Approvals.” The censored profanity is black-barred, while the final word “Code” is left visible, setting up the gag. The scene lampoons the lengths a developer will go to obtain mandatory pull-request approvals, highlighting pain points of code-review bottlenecks and the sometimes transactional feel of the process
Use J and K for navigation
Comments
6Comment deleted
Nothing says ‘modern CI/CD’ like bartering roadside favors for an LGTM because the CODEOWNERS file still thinks 12 ex-employees need to approve your one-line fix
After 15 years in the industry, you realize the hardest distributed systems problem isn't consensus algorithms or CAP theorem - it's getting two senior engineers to agree on whether that variable should be camelCase or snake_case in your PR that's been sitting there for three weeks
Every senior engineer knows that moment when your carefully crafted PR sits in review purgatory for days while you've already context-switched to three other features. You start pinging reviewers on Slack, dropping hints in standup, maybe even offering to review their PRs first - anything to break the deadlock. The real tragedy? Your code is blocking the entire sprint, but Bob from Platform is on PTO and he's the only one who understands the legacy authentication layer you touched. Meanwhile, your feature branch has diverged 47 commits from main, and you're one force-push away from a nervous breakdown
PR approvals are the new OAuth tokens - you burn half your day acquiring them, they expire in the merge queue, and the scopes never quite include "allow merge"
When branch protection demands two reviewers, a man's gotta barter LGTMs like scarce compute credits - because the real bottleneck in CI/CD is human attention
Mastered CAP theorem, but can't CAPture that one reviewer's approval without selling your soul