Everything OK, Github?

:date: 2015-03-12 15:08 :tags: image,git

A few days ago, I wrote about my thoughts on Git and Github. Today I had a reason to add another point to consider. Looking for some software which was hosted on Github, I got an HTTP 500 (500 Internal Server Error) with a cute animated cartoon saying that "something" was wrong.

The problem was fixed in a few minutes but I got to imagine what life would be like if all my code was stored there and such an event was more protracted. I think that Github is pretty sane and if you think you should use it, you definitely should! But for those of us who don't, remember, we have our reasons.

Also, ever wonder how long your data will be reliably hosted on Github? No one knows for sure of course, but Google Code might give you a hint. Looks like about 5 years. Good luck!

UPDATE: 2016-01-27

Here it is again.

My software repos had better uptime than this in the last year.

UPDATE: 2018-02-12

Here's an interesting article pointing out the subtle problem with allowing (other) people to reclaim deleted GitHub user accounts. Using GitHub to spread copies of source code far and wide can't hurt, but banking on critical runtime hard links to specific developers' accounts should probably be avoided.

UPDATE: 2018-05-29

Did the little Github pentopus just literally take a dump? Well, figuratively it definitely did for me today.

UPDATE: 2019-01-24

WTF Github! Stop stealing important browser keyboard bindings! Now in Github when you hit "/", God's correct key binding for "search", Github hijacks that and moves focus to their own predatory bullshit remember-everything-you-ever-searched-for search. Gaaah!! I hate this!! NO!!!

Here is a good explanation of why they do this. And it is not good enough!!!.

UPDATE: 2025-02-27

I think we could all see this one coming. TechCrunch reports Thousands of exposed GitHub repositories, now private, can still be accessed through Copilot. I think there are three obvious lessons from the article. 1. Anyone who stores keys on Github or anywhere else not completely in their control is a fool. 2. Anything ever made public for any length of time on the internet must be considered replicated and distributed to everyone on earth, forever. There is nothing really new here as this also applies to archive.org and search engine crawls, etc. 3. Although a bit contrived, this article describes a real case where the new LLM AI tools are disclosing private data that found their way into their training corpus; this is a thing and it makes sense to keep private IP out of any possible training data source. Like Github.