📣 Opinions are welcome 📣

[erikjan@fosstodon.org]

@Gina I want exactly this for scientific code!! Maybe at eu level, with a nice url like code.science.eu

Don't want to rely so much on GitHub anymore

[gina@fosstodon.org]

@xiu that was my argument as well. The threat when self-hosting is no more updates.

(In case it isn't clear I'm strongly pro Forgejo, but as GitLab is already popular with our gov orgs it needs to be considered - which I'm aware also isn't the best argument)

[gina@fosstodon.org]

@erikjan yeah I could imagine an EU federation of code platforms at some point 🔥

[jbouter@fosstodon.org]

@einalex @Gina true, but that might be quite a few changes. There are hard requirements by the ADR that I know of for example. One of them is a mandatory 4-eye principle on code going to production. Forgejo has support for codeowners, but no option to disallow a single codeowner to push to prod.

[gina@fosstodon.org]

@jbouter @einalex yeah this is also part of what the pilot should focus on. All the rules, regulations and technical demands that a platform should be able to comply with. 🥲

[owlcode@fosstodon.org]

@Gina a big institution,like a nation, as a user could push the development and fixing of the platform at an important level

[gina@fosstodon.org]

@owlcode absolutely, and finance (think of what GitLab EE licenses cost)

[jbouter@fosstodon.org]

@Gina it is... a lot! I've been part of a few implementations and it's never easy. Both on functional and technical demand. Best of luck figuring this thing out! I look forward to working on a govt-wide git forge someday! 😁 it might be worth checking in with ODC-Noord. I think they already allow government agencies to sign up to their git environment. How they came to certain design decisions and such ☺️

[fedops@fosstodon.org]

@Gina same. It's mostly a timing issue in many cases. If you need the functionality now you may have to pay for it. If you can contract with someone to implement it for you, say over 6-12 months then everyone profits.

[jspijker@fosstodon.org]

@Gina Does it matter? At our gov institute we have an internal/private Gitlab instance, doing all the CI/CD for our (very specific) pipelines. For our public/open projects we use Github, with some limited CI/CD (mostly generating Github pages). Both Gitlab and Forgejo can do that. From a user perspective, I would be very happy with a gov owned codeplatform with basic features. For more complex stuff you can roll out an internal instance with all the needed features.

[gina@fosstodon.org]

@jspijker exactly, in which case Forgejo would be a great option. For anything more complicated an org can always set up their own GitLab (or whatever they like).

[smlx@fosstodon.org]

@Gina @jbouter Gitlab EE also has astronomical licensing costs at scale. It might be useful to consider the cost over a multi year period vs funding the additional features you need in Forgejo.

[jbouter@fosstodon.org]

@smlx @Gina yep, very true indeed. It depends very much on when you need those features and how important that mandatory support contract is

[gina@fosstodon.org]

@jbouter @smlx the support contracts are def a legal requirement (usually a bottleneck for gov procurement of opensource solutions), but if structured well we might be able to solve that with a service provider in between.

[erikjan@fosstodon.org]

@Gina ok, then nl.code.eu & science.code.eu 🤓

[ianthetechie@fosstodon.org]

@Gina voted for GitLab mostly because I have direct experience with it and I have none with Forgejo. Also I could be wrong but it seems to be more mature.

It’s really a hard question though. One is backed by a company which could always turn. But it’s also a lot more likely to be around and actively developed in 15 years (IMO).

I’ve self hosted GitLab CE (and more recently upgraded to EE for one feature which I don’t remember) for over 10 years and have no complaints.

[gina@fosstodon.org]

@erikjan btw there's a good chance that @SURF is already working on something like this (and if not, they would be the first I would think of) @wlaatje

[gord1i@fosstodon.org]

@Gina we've been using GitLab community edition internally at the City of #CapeTown for about 7 years, and it's worked well for us, without any of the paid for features.

I think a big question is how much of the project management stuff is core vs *nice to have* - do you need just git + a code review platform, vs git + code reviews + an internal Jira, too?

I've been fairly surprised how effective/under developed the idea of issue management has been in our context, so GL has worked for that

[aaronmccollum@fosstodon.org]

@Gina Voted Gitlab, no issues with it personally and it's more well-known. Both look like good options though.

[suetanvil@freeradical.zone]

@Gina

Why not use (and fund) Codeberg? It's already established and has a user base and it's located in the EU.