Today, I'm making a revolutionary change to our code base. I'm finally deprecating a script that lived in this goddamn repo for way too long. There's about 10 copies of the same script in 10 different directories. The script copies all the code into a "dev" folder and then runs "sed s/prod/dev/g" on all the files, and then overwites a bunch of it with some files suffixed with ".dev". Finally, after fighting the so-called architects, the devs and everyone else that seems to have gotten used to the pain this cursed dumpster fire script has caused us, my merge request is now open and ready to go to get rid of this insanity. Now we won't have to deal with as many "surprises" that happen every goddamn time we deploy to production, overwriting all our hard work by accident, and relieve some of my OCD of having the same script in 10 different places in the repository.

  • 0
    God damn time bomb.
  • 3
    @N00bPancakes it was a reusable time bomb, it exploded multiple times already lol
  • 0
    But but but... What will the insane ppl blame now???

    It is just like my work:
    Each project (should be a monorepo) has a dependency(!) to a Makefile(!!) project - that holds all the build(!!!) scripts in it. It was "fun times" trying to make the builds reproducible.

    Makes the whole "but it works on my machine" conversation more interesting.
  • 0
    @magicMirror people were running a bunch of shell scripts to deploy to prod from both mac and Linux, which of course doesn't behave exactly the same... I just found out that we have a "deployment server" which kind of helps some of those issues, but it's set up so terribly. It's LDAP auth via SSH and would map your LDAP groups to your user account, and gives you access to the cloud environment through a service account based on your groups.. But everyone also has sudo so they can just su to a user with higher privileges and do whatever they want in prod
Add Comment