Looking for a better way to #git.
Tired of #libre sofware #sourcecode being centralised on big players' platforms.
Sourceforge is dying a slow death. Github is too big (but not too big to fail!) Gitlab is now controlled by VCs.
I could setup my own git server, but don't want to isolate myself from potential collaborators.
Today I tried this 3 repo setup:
- Main repo on https://framagit.org by @Framasoft
- Gitlab.com setup as a mirror bridge, pulling from Framagit and pushing to Github
Sounds like someone needs to develop gittorrent. Probably could build on btsync and avoid reinventing the wheel.
The crazy thing is that git is already federated. It was designed that way from the beginning. So once you clone a repo, you can set multiple upstreams and then push/pull to your heart's content (modulo permissions).
Discoverability is another thing, but does not have to be tied to git itself.
I have local copies of things with upstreams all over, GitHub, gitlab, git-ssb, repo on Dropbox folder, repo on filesystem.
Yeah, it's a tragedy. Even when using decentralised systems we tend to converge (centralise) on the same services anyway (eg. Gmail)...
@mayel @alanz @cykros @Framasoft I've recently found out about "Anti-Disintermediatization" where formerly decentralized tools were centralize-able, and hence were centralized. Kleiner suggest we make tools which *cannot* be centralized, where they only run on your local machine.
Which (IMO) is a fascinating new perspective (for me)
@ebel @Framasoft @cykros @alanz @mayel
We did all of this stuff for long enough before the silos came along, we used just mailing lists and usenet.
One of the reasons RMS doesn't like APIs as I recall is that they allow non-free software to interact with free software, where the preferred alternative is that anything leveraging your code basically has to be a derivative work.
@gracie @ebel @Framasoft @cykros @alanz @mayel
Part of the problem is that modern software development is more than just posting code somewhere. You really want to have some kind of forum, an issue tracker, a wiki, integrations with build systems etc. etc. It's annoying to have to set up, run and monitor each of those things piecemeal. Similarly, it's often unrealistic to tell people to run their own instances of X. Not everyone wants that kind of responsibility.
@gracie @basus @mayel @alanz @cykros @Framasoft @ebel Does it really? I don't participate in the Linux kernel list but all the other mailing lists I attempted to participate in left me thinking that there was no way to produce good work in that environment.
So much duplicated reading and writing, and constant side-tracking of conversations causes things to progress much more slowly than necessary.