Based on this recent post of prot to leave sourcehut I thought it could be interesting to start a discussion about hosting a git sever for the system crafter community, for example with forgejo.
Are there already any plans for this?
Of course this should host the content directly related to system crafters, e.g. crafted emacs or crafted guix.
What else?
It could also host new community projects.
It could mirror emacs configurations that could be used as example configs for others.
Same for guix.
It could also mirror guix channels. So far they are all over the place (on Github, Gitlab, SourceHut, CodebergâŠ). That way it would be easier to find packages for guix.
On a meta level, it could even work as an entry point for new members to contribute to software that uses âalternativeâ code hosting platforms (FSFâŠ)?
I am of course aware that it would not be for free, but I guess the benefits are well worth it.
I have considered it a few times but wasnât sure whether people would find it worthwhile.
You do raise some interesting use cases though regarding mirroring repos. The SC Git doesnât necessarily need to be the primary home for all repos, but it could be a place where repos can be aggregated to be found more easily.
Anyone else have thoughts or suggestions about this?
The mirror sounds like a good idea, especially since you cannot deactivate the issue tracker and merge/pull requests on some forges. Thatâs why GitHub - torvalds/linux: Linux kernel source tree still has pull requests open, you cannot deactivate them, even though Linus will never use those.
Iâd generally recommend against it, unless you really have the desire to provide a long-term home for peopleâs software projects. There are already so many good forge hosts out there. A list of the top-3 SC-recommended ones could be as good as or better than hosting one yourself.
Hmm. I see the point against hosting software projects directly, considering that nobody knows the long-term outlook of this community.
But I still think my use case of mirroring certain projects is valid.
It happened two times for me now that I was working on package definitions for Guix only to find out later someone else already worked on them. Both times I only found them more or less by chance via the mail archive, because the indexing of e.g. Gitlab is so bad.
Similar for Emacs configurations, that I only found because someone linked them on e.g. Reddit.
Also, itâs just sad imo if even Prot is now mainly relying on Github (as do most other Emacs packages).
Fair point! I wouldnât want to do it unless I knew I could provide stable service long term. The ideal solution would be to pay for Forgejo hosting from someone reputable so that I donât have to deal with outages.
Iâd say the likelihood of it happening is like 60%.
It seems to me that the way to discover Guix package definitions ought to be through an index provided by Guix (e.g. for non-Guix definitions and channels, there could be a community index on the Guix site). Searching forges or googles wonât ever be a reliable way to find them.
Maybe instead of a full-blown forge, just a curated list of System Crafterâs projects? Something like a SystemCrafters/awesome-projects repository, similar to the other "awesome" link lists on GitHub and other places?
