Rolling, rolling, rolling back nothing I love more than communicating why we had to roll back again (ノಠ益ಠ)ノ
You guys get to test?
Question 1: Where’s the CI/CD pipeline?
Question 2: Why can one person change production alone without peer review (outside of an emergency)?You don’t have a job, you have a ticking time bomb.
It’s internal development (Portugal) developing something for an internal department (Germany). There’s nothing professional going on here so we are far from any ci/cd pipeline. One person can change everything, because it’s just two developers (1 frontend, 1 backend).
Plus things are busy and we (team in Germany) are way more interested in this thing working well than they (team in Portugal) are since they have higher priority tasks…
Of course I can say fuck it and live with the poor quality caused by circumstances that were partially caused by poor management decisions of the company but I’m not able to care little enough.
I demand CI/CD for my scripts. If it runs in prod or against prod or anywhere near prod, it gets a pipeline.
Technical maturity isn’t just for big companies and important things. It’s a practice. Why half-ass something when you could whole ass it?
Pre-commit hooks don’t require a pipeline nor any money. In most cases it’s one line of code to make the tests run every commit
Even better: use pre-commit. It supports all kinds of stuff without a lot of config. This gets you (and GP) a lot of the features of a full-blown CI pipeline, but it all runs locally before anyone breaks anything.
Stop me from committing my work and I will hunt you down to the ends of the earth.
Fix your shit and it won’t stop you from committing.
It’s also usually only on certain branches, so you can make a branch where you break things and then fix them before you merge to testing/main/whatever.
TIL precommit hooks can be set per branch. I was being facetious to begin with but this sounds pretty good actually.
Nah, at our place it’s applied on all branches…
What do you do if you have code that isn’t complete enough to work? Do you have to just leave it untracked?
I don’t know what others do, but I personally whip out git commit -n and bypass the hooks in this situation.
If you have code that is not complete it is not qualified to be deployed. Cut work items into smaller chunks but never deploy not fully, 100% working and tested stuff. Not even on dev.
Take down prod while I’m on call and seeing my kid and I shall return the favor
I get that a lot
Getting threats over one line of code is called senior development
Don’t worry I’m too lazy to hunt you down farther than the coffee shop next to me.
I agree. I absolutely hate when some pesky git hook rejects some debug code I wrote that I want to commit. Mind you, commit, not integrate. This is the situation where I whip out git commit -n.
If you have tests you have to automate them and have a pipeline which stops any deployments when these tests fail. You definitely have a mindset issue and not a management or tool issue. I am not sorry but as a DevOps Engineer I can only say you guys are a lousy hack and should probably seek a different job.
It doesn’t have to be that way, if you have a CI/CD process that prevents it.
Why kind of psychos are merging with failing tests?
The kinds of psychos that work at companies that measure performance by number of commits and PRs, successful or not.
That explains why vibe coding is so popular… just commit after every iteration the LLM spits out
The kind of psychos that have “nice to have” tests. If it’s red, it’s not critical, but still worthy of attention… sometime.
the real question is why doesn’t the CI/CD prevent merging with failing tests in the first place. i have not worked in any company that allowed that for at least 15 years.
Rn I’m in a project where everyone that has access to the code is given the role of owner of the group so we have permissions to skip any and all measures since owners don’t care.
I’m so happy that tomorrow is my last day. So happy.
haha, i can imagine :)
Proper procedure would have been the dev updating the uts so that they pass. If all they updated was adding ignore or commenting out code in the ut, even better!
because they know better than your tests obv.
