microsoft buys github

[sgrogan]

I already registered FreeCAD on GitLab maybe a week or two ago. I run my own self-hosted GitLab server with integrated login from FreeIPA (<3) and there's a possibility of extending your own self-hosted instance with GitLab CI workers, so I was looking into that as a possible alternative to the Buildbot portion of my GSoC stuff.

Can we mirror FreeCAD now? We had a GitHub mirror when we were still on SourceForge.

[freedman]

If I had to guess what they will do; They won't mess with Freecad, there are too many people watching. They will start off by targeting the smaller rarely used stuff and (once they find there niche) move to larger programs with more users. By then they can say " we have been do that for 2 years".
It's the same old thing, start small and take just a little, and then a little more. After enough time has passed, it's called "Normal operation" and no-one can argue with normal operation.

Then the big hit comes when the word SECURITY is yelled out. Everything is justified at that point.

OK!, I just kicked my soap-box out from under me. EVERYTHING WILL BE FINE.

[kkremitzki]

If I had to guess what they will do; They won't mess with Freecad, there are too many people watching. They will start off by targeting the smaller rarely used stuff and (once they find there niche) move to larger programs with more users. By then they can say " we have been do that for 2 years".
It's the same old thing, start small and take just a little, and then a little more. After enough time has passed, it's called "Normal operation" and no-one can argue with normal operation.

Then the big hit comes when the word SECURITY is yelled out. Everything is justified at that point.

OK!, I just kicked my soap-box out from under me. EVERYTHING WILL BE FINE.

In my opinion they won't try anything even that direct like Sourceforge. If you look at MS's current modus operandi, they are interested in data - Skype, LinkedIn, Windows 10... And now with Github. Imagine you want to make an AI that can write code or just help programmers with code. What dataset would you want to train it on? Exactly the stuff Github has. The value proposition of something that could do that is so huge, you just might pay $7.5 billion dollars for it.

[kkremitzki]

I already registered FreeCAD on GitLab maybe a week or two ago. I run my own self-hosted GitLab server with integrated login from FreeIPA (<3) and there's a possibility of extending your own self-hosted instance with GitLab CI workers, so I was looking into that as a possible alternative to the Buildbot portion of my GSoC stuff.

Can we mirror FreeCAD now? We had a GitHub mirror when we were still on SourceForge.

Yep, I also need to do a GSoC update post tonight but this is on my list as well as adding people to the GitLab FreeCAD org as appropriate, i.e. the people in the Github FreeCAD org.

[freedman]

Imagine you want to make an AI that can write code or just help programmers with code. What dataset would you want to train it on? Exactly the stuff Github has. The value proposition of something that could do that is so huge, you just might pay $7.5 billion dollars for it.

Interesting... So now I have to start asking; Hey, Kkremitzki are you AI? You seem way too smart to be human. How's that for a compliment.

Just kidding if anything is taken the wrong way.

[kkremitzki]

Imagine you want to make an AI that can write code or just help programmers with code. What dataset would you want to train it on? Exactly the stuff Github has. The value proposition of something that could do that is so huge, you just might pay $7.5 billion dollars for it.

Interesting... So now I have start asking; Hey, Kkremitzki are you AI? You seem way too smart to be human. How's that for a compliment.

Just kidding if anything is taken the wrong way.

Hah, thank you.

ping

We now have https://gitlab.com/freecad/freecad. It's importing ATM... and I've added you as an owner for the FreeCAD group. I imported copies of all the github.com/freecad repos into gitlab.com/freecad.

[yorik]

Thanks Kurt, you're taking the bull by the horns, as we say in french.
Can gitlab keep things in sync automatically? I suppose not, but it's fairly easy that someone adds a commit hook to it, I do that with my personal github repo already..

Even if microsoft goes the bad way with github, this will be a very easy move out, for us as for any other project hosted there, remember the sourceforge suicide. Indeed something like QT being bought would be a bigger problem. But we would be together with a large part of the open source projects of the world, I guess an "openQT" fork would appear the same day and most of the non-corporate QT developers would migrate to it. OpenCasCade would be a much more annoying problem, but there also, they learned to not underestimate the community, their project was almost forked out of their control at some point (the PythonOCC/OCE glory days), the same could happen again. You cannot unlicence what has already been open-sourced.

If microsoft has regained any little bit of notion, they will be VERY careful. This will be interesting to watch, from an anthropological point of view

[keithsloan52]

If I had to guess what they will do; They won't mess with Freecad, there are too many people watching. They will start off by targeting the smaller rarely used stuff and (once they find there niche) move to larger programs with more users. By then they can say " we have been do that for 2 years".
It's the same old thing, start small and take just a little, and then a little more. After enough time has passed, it's called "Normal operation" and no-one can argue with normal operation.

Then the big hit comes when the word SECURITY is yelled out. Everything is justified at that point.

OK!, I just kicked my soap-box out from under me. EVERYTHING WILL BE FINE.

In my opinion they won't try anything even that direct like Sourceforge. If you look at MS's current modus operandi, they are interested in data - Skype, LinkedIn, Windows 10... And now with Github. Imagine you want to make an AI that can write code or just help programmers with code. What dataset would you want to train it on? Exactly the stuff Github has. The value proposition of something that could do that is so huge, you just might pay $7.5 billion dollars for it.

Like the thinking, but FOSS software like FreeCAD is in the public domain and freely accessible without buying GitHub.

I think the guys that have private paid for Github repositories would have a bigger issue as it looks like Microsoft will now have access to their source. I know IBM now uses github, wonder what they think of the deal?

[kkremitzki]

Like the thinking, but FOSS software like FreeCAD is in the public domain and freely accessible without buying GitHub.

I think the big gain for them is not the code of any particular repo (or all of them), but the metadata around it, like comments in issues and pull requests which discusses code and has code changes as results of those discussions, having data about individual developers who work on multiple language projects so they can extract the "developer" signal out of "developer + code" noise, and developer preferences like stars and follows. While a lot of that stuff is public, you could hardly just web scrape it off Github without getting your API access rate limited and charged out the wazoo.

[keithsloan52]

While a lot of that stuff is public, you could hardly just web scrape it off Github without getting your API access rate limited and charged out the wazoo.

Try googling for "scan github for secrets" you will get lots of hits for tools like trufflehog that scan repository.
And articles like https://www.zdnet.com/article/truffleho ... he-masses/
Did not get the impression that users would end up being charged loads of dosh.

Apparently, Amazon scan github repositories looking for things that expose secrets that would allow people to hijack users services