GitHub alternatives

Those of us that remember the world before GitHub (it only came onto the internet scene in 2008), yes, it was like a breathe of fresh air at the time and it really did help to cement having a "free" hosted environment accessible from anywhere where you could store your code, share the code, manage the code and collaborate with others with your code.  It was refreshing.  Before then you had tooling, like CVS that, if you were lucky enough, you could install locally inside a company and have 1001 variant modifications that meant no CVS was the same as another, so headaches ensured.  I'm sure there were other tools, I seem to recall doing stuff with Rational Tooling too.

(We'll come back to these people later on!)

So what happened?  Well, as with all good things, they attract attention.  The original creators also tail off & then look for different and other fun things to do and well, usually somebody in a mega-corp then decides they want to buy the system / service / software (MySQL anyone?!) to mainly get to the user base, but sometimes to add into their own strategy.  Usually, they end up screwing it up and alternatives are born and a lot of people jump ship, some stay as they don't want the hassle of setting stuff up again, they are comfortable. (MariaDB anyone?!)

And the same is true of GitHub.  Microsoft decided, back in 2018, that it would help them significantly if they had a major slice of this pie:

So what, Tony? So what?

Well, I do recall doing some research at the time about Natural Language Processing (NLP) and understanding code and how to create new code based upon Deep Machine Learning (ML) so that you could generate new code without being experts.  However, you need a ton of example data of existing code bases to make this work.  Oh and there is a tool with the worlds largest repository of code, just there... GitHub.

I remember finding Microsoft Research papers at the time that indicated the direction Microsoft Research was going in, so I was a little sceptical about using GitHub.  The "free" version allowed you to create projects and check in/out your code, but if you wanted it "private", naturally you had to pay.  However, if you were the company owning where all that code was going to be stored, those definitions didn't matter.

What a perfect setup - you would have access to a massive repo of source code that you could then use to teach a machine learning model how to code.  Well, I imagine that there still needed to be humans in that loop as code is code, it is subjective, is it "good code", "mediocre code" or just "bad code"? that is an opinion.  does it work? well, yes, it might.  is it the best way to code for that scenario? maybe not.  but hey, there are 1m examples to choose from, so your algorithm could work out what is deemed as good or bad.

here's an article about Project O with a link to the Intel white paper: https://tonyisageek.blogspot.com/2020/12/project-o-my-machine-programming-project.html

I do recall I printed out the Microsoft Research white paper, but am failing to find the link, it was 7 years ago!!?!?!

Well, you get the point, Microsoft was lining itself up, way before OpenAI released anything to do with chatGPT, it's like certain companies or individuals were working behind the scenes on a future strategy or something.  LOL.  anyway, so what have Microsoft done with GitHub? Did they leave it alone?

Nope, they decided to go all-in with CoPilot.  They decided to integrate CoPilot with VSCode - I admit they did a great job of pushing VSCode out to the masses, even converting Linux developers to start using as an IDE.  I think I even have/had it installed, mainly because I got pressured into it as someone wanted to use my laptop and Python and they'd only ever done that using VSCode, so "I" got VScode-ified.

Roll forward to today / well, now-ish.  People are starting to have a little bit of a backlash against the "AI" tooling (CoPilot in this case) from fiddling with their code, they don't want it, they can do without it.  They uninstall it, it re-appears, it comes in a different way.

Why? Well, Microsoft dropped Billions of dollars into this CoPilot equation - they did the same as what IBM did a few years back, they gambled everything on "33 Red".  So now they are rigging the roulette wheel to go around a bit more to force the ball to land in "33 Red", or on "Red", they cannot fail.  That would be bad.  It would be an admission that they'd been pushing the hype.  I get the concept, I was there in the eCommerce / Internet bubble bursting days back in 2001/2002, people were sold on the hype and the investors were pumping money into everything with the hope that by the time it comes around to realisation, "someone" would have figured it all out and made the hype a reality.  Timing is everything.  Unfortunately, the recent "AI" bubble was leaked out to the general public, it got them engaged, hoping to then get an inside influence into businesses; however, that kind of hasn't happened, yes, if you go look on YouTube, you'll see 10000s of videos about using AI to make millions, automate workflows, do loads of things simpler & easier and there are 20 year olds sitting around swimming pools in Bali claiming they are earning $10k per month from "AI" stuff, who knows they might be, but it's not sustainable.  It will burst.  And that is what Microsoft is going to feel, but on a larger scale.

Oh here's an ElReg article about this, so its not just me or my one sided opinion :-D

Okay, so now you have context, let me introduce to you a GitHub alternative:

CODEBERG

https://codeberg.org/

and that is actually powered under the bonnet by one of the repo's that it hosts:

FORGEJO

https://codeberg.org/forgejo/forgejo

https://forgejo.org/

As the website states, you can download and use this tooling locally or host it yourself, with no intrusions, just to do what it says on the tin.  It'll run on a Raspberry Pi more than comfortably!  So if you're worried that you'd need a mega-server to host this on, nope, just stick it on a cheap old RPi4 that you've got in the drawer not doing anything, hook it up to the WiFi that you're always connected to and job done, it's there, it's doing it's thing for and with you.  No forced "AI" BS.

Download for docker to run on a linux machine: https://forgejo.org/download/#container-image

Then follow the installation docs here: https://forgejo.org/docs/next/admin/installation/

and YES, I also didn't miss the irony about referencing docker just then - they also have instructions on using PODMAN too :-D

Talking of PODMAN instead of DOCKER - here is a super relevant article:

https://codesmash.dev/why-i-ditched-docker-for-podman-and-you-should-too

you guess it - yep, he shows some of the security technical "flaws" of docker and basically convinces himself and probably you too(?) to switch over to using PODMAN.

I'm just about to set it up locally myself, whilst DOCKER was good for "Personal Usage" then they started charging for the "Enterprise Usage" which makes sense from their business model approach, but y'know.... like GitHub, it's become "corporate" and a CAPITALIST tool.  Time to switch to the RedHat version.

Think it's "too difficult"? think again:

Right, I'm bored now, so I'm off for a walk & do something "outside" that is not a laptop screen or technology focused.