I understand it's a joke, but really isnt the entire point of git is to be able to work locally as much as you want without affecting the remote repo and vice versa
Git allows me to write code as much as I want. But GitHub does more than just Git. If you don’t remember the details of the next task you need to work on and GitHub is down, that’s a problem. As a senior I spend a lot of time reviewing PRs. That’s considerably harder when GitHub is down.
I mean there are tons of options in that space so if it's an issue that is sorta on your business to have evaluated their dependency.
We work on an internal gitlab instance that has had 100 percent up time for like 2 years. It doesn't even have to be gitlab, there's gitea and like 10 other options.
I personally think that the industry has moved so far in the direction of cloud and saas that it's lost a lot of valuable skills and made them dependent on too much externally.
It's like "nobody ever got fired for buying IBM." Nobody ever got fired for pitching a migration to GitHub. It doesn't have to be good. Then one day it's crumbling down and people will have to learn to face consequences.
I'm the only person at my (small startup) company who has the skills to maintain a GitLab instance. Been there, done that, never fucking again. I HATE maintenance. We're probably going to migrate to some other platform since GitHub is intent on turning to shit.
In 2014 I set up GitLab for my then employer. It had to be something self hosted because of client requirements. I was apparently the only one in a company of about 200 that knew anything about Linux.
Wasn't too bad, just keeping it up to date etc. When I left in 2016 I'd just upgraded the server to ubuntu 16.04. It's probably still running that now. I know someone who is still there and they've said GitLab itself hasn't been updated since I left.
I set up and maintained a GitLab instance and GitLab CI runners for five years. It was fine. I still hated it. I loath maintaining infrastructure.
To each their own but ours didn't really require more than an hour a month at most. It's not running on cutting Edge hardware but chugs along pretty dependably. The back ups probably take the most time but even then ansible does most of the work and we bump the omnibus version once a month in off hours without issue.
It’s not as much time as it is stress, anxiety, and trauma. Being on call when shit breaks is fucking awful and my best coping strategy to date is refusing to be an infrastructure person and aggressively not giving a fuck when things are down for a day or two.
Sounds dumb to be that dependent on a US platform in 2026 AD
Right? We’ve had two thousand and twenty six years since Christ walked the earth to reduce our dependency on GitHub, what are we even doing
Everyone knows we were banished from the garden of forejo after Steve made the Apple and even Jesus dying wasn't enough to let us go back.
What do you use for project management? What platform-less system are you using for that? Or are you saying to use a non-US platform? Do you have specifics.
Well couldn't you have your own self hosted "Git(hub)"?
which is absolutely true until you wire your CI pipeline through it. Now it's a critical fucking deploy function for dev/stage/QA and maybe prod now with workflows.
You can run the pipelines locally. But it's complicated so it's better have your own scripts and keep the pipeline short.
There are more ways to set up a pipeline than there are ways to shuffle a deck of cards.
Everybody seems to like to tie back into online services. People like github workflows, and using NPMs and external DNS and docker Deps and JFrog. By the time you chain all those SLAs together you've got a bucket of risk the size of a small bus.
I try to push them as much as possible to use straight up bash scripts, and then call those with automation.
If it were solely up to me, I'd host my own repositories, but at some point, risk and safety end up losing out to some extent to features and feasibility.
For some reason tons of developers moved that amazing concept to depend as much on Microsoft cloud as possible for their workflows.
It's possible to design your devex to require an unreliable SAAS vendor for even basic tasks! If you try hard enough you can logjam your entire team!
GitHub has actions etc. a lot of people don’t build locally. They push to GitHub and it builds, tests, deployed, does checks etc
You should be able to replicate at least some of that locally. If you can't work with GitHub down for a couple of hours, then it's a poorly set up project.
"Please Microsoft, I beg you. My penis is starting to get sores from all the sex I am having due to being unable to do any work."
and everyone else at the office is complaining.
I don’t really get this joke as I’m not a developer but does it have something to do with that thing where I try to search the site and it tells me it’s getting too many requests from my IP, even though I haven’t searched it in a month?
GitHub is where a lot of companies store their source code, so many software development workflows require it to be available. For a while it had fantastic uptime, but since Microslop started shoving vibe coded updates its reliability has cratered.
GitHub manages not one, but TWO nines of availability.
Sometimes both nines are even in the front!
It's been Windows 11'd and the regular patches of downtime are just one of several new productivity-loss features to be rolled out.
Next; ads.
...no, wait, that's already happened.
Next; sponsors.
...actually, wait, that was kind of done.
Next; a bloat UI front-end to minimise the confusing layout. Subscription to opt out.
Nah, next is replacing all UI with a copilot prompt that always shits on your code and guilt-trips you when you try to look up some repo instead of just asking it to vibecode it for you.
codeberg was down a few times in the past days, but i feel like its got more uptime than github atp
At least codeberg isn't a billion dollar company
My self hosted gitlab instance has better uptime over the last 12 months than a billion-dollar SAAS product 🤣
It's reliability was already not super great before the vibe coding hit.
It's mostly caused by them trying to migrate complicated legacy systems to azure. This type of work is already fraught with danger.
The vibe code just supercharges an already risky endeavor. The LLM code could be perfectly correct but the more dynamic a complicated system the more difficult it becomes to judge side effects of any one change.
The speed of change also means that experts in particular areas of the code base may find their mental model of the system to be out of date and incorrect faster than ever before. This is course also leads to the increased chance of mistakes or unintended side effects.
Luckily we use gitlab instead of github
It also has some downtime now and then but it isn't owned by microslop which more than makes up for it
I'm still trying to work out how to do ci tests without GitHub actions or a credit card or self hosting.
The gitlab pipeline stuff is pretty good, and the free plan comes with a decent amount of compute, i assume no credit card is needed though i haven't used it in any private projects so i'm not certain
Use Codeberg! Mostly up, immediately filters AI grifters, and isn't tied to Micro$lop.
work
codeberg
Pretty sure they don't allow private repos. Great for open source projects though.
They have private repos, they're paid though.
There are other Forgejo hosts though, and of course you can host your own. Codefloe allows private repos and has a free tier, but apparently has paid tiers for when you want to use more resources
bookmarked!
Also moving all my stuff to codeberg.org
Anyone got a good graph summarizing this decline over the last few years? Kind of want to show coworkers
I wrote mobile apps for Blackberry back in the day. As part of their security fixation, all library modules you incorporated had to be signed as your app was compiling, even if you were just testing out a single line change. This could make your app take upwards of a whole hour to sign, if the signing servers were even up and running at all; they were often down completely which meant I could go home and get high instead of working. Which is why I never badmouthed Blackberry to my bosses.
The absurdity of having every module signed meant that I had to think long and hard about whether I wanted to use built-in library functionality or just roll my own code. For one UI I needed to use trigonometry functions. These were located (logically or not) in one of the encryption modules which were especially prone to taking a long time to sign, so I ended up writing my own sin()function (in Java) just to save myself ten minutes of compilation time.
GitHub being up really does feel like a limited-time event now. Better push while the servers are feeling generous.
Welcome to Programmer Humor!
This is a place where you can post jokes, memes, humor, etc. related to programming!
For sharing awful code theres also Programming Horror.
